You are on page 1of 27

REVIEW

Thermoelectrics
www.advelectronicmat.de

A Brief Review on Measuring Methods of Thermal


Conductivity of Organic and Hybrid Thermoelectric
Materials
Hanfu Wang,* Weiguo Chu,* and Guangming Chen*

measurement methods and to invent new


Organic and hybrid thermoelectric (OHT) materials have attracted increasing strategies to fulfill the needs of character-
research interest over the past decade. Thermal conductivity measurement izing diverse samples ranging from bulk
plays a critical role in evaluating and improving the thermoelectric perfor- materials to low dimensional structures.[7,8]
For the three properties that are used to
mance of various types of these novel materials, ranging from bulks to low- determine the ZT value, the characteriza-
dimensional structures. Commonly used and newly developed techniques tion of the thermal conductivity usually
for thermal conductivity measurement, most of which have been applied to poses a substantial challenge. This is mainly
OHT materials in recent years, are reviewed. They are categorized as steady- due to the fact that unwanted heat losses
state methods, time-domain methods, or frequency-domain methods. The (e.g., from blackbody radiation) and thermal
contact resistance (e.g., from the interface
operating principles, merits and limitations, technical issues, and application
between the sample and the temperature
examples for each technique are also discussed. sensor) can more or less bring extra errors
to the final results. Measures must be taken
to reduce or to quantify these influences.
1. Introduction The OHT materials generally share same characterization
methods of the thermal conductivity with their inorganic
In recent years, organic and hybrid thermoelectric (OHT) mate- counterparts, except that the measurement temperature for
rials have attracted tremendous attention due to their easy the OHT materials is usually lower than 473 K. However,
process ability, low cost, and low thermal conductivity.[1–6] The special attention should be paid when applying these tech-
thermoelectric performance is generally described by a dimen- niques to the OHT materials due to the fact that the meas-
sionless thermoelectric figure of merit ZT urement accuracy and reliability are significantly affected by
their unique mechanical, morphological, thermal, and optical
S 2σ T properties. For example, the relatively large thermal expan-
ZT = (1)
κ sion coefficient associated with the polymer materials[9] can
cause thermal contact problem in the steady-state methods,[10]
where S, σ, and κ are the Seebeck coefficient, electrical con- especially when a large temperature difference is exerted. It
ductivity, and thermal conductivity, respectively; T is the tem- is also known that certain polymers and polymer composites
perature. A good thermoelectric material should possess a high may become mechanically unstable at cryogenic tempera-
power factor (S2σ) and a low thermal conductivity. Reliable tures.[11] Hence, one should make sure that measurements at
evaluation of these properties with proper characterizing tech- very low temperatures do not cause the samples to deteriorate
niques is crucial for developing advanced thermoelectric mate- physically (i.e., generating cracks or fallen pieces[12]). Surface
rials. Tremendous efforts have been paid to improve the existing roughness of the OHT materials is another concern, which
is usually not easy to control in the fabrication step. It may
eventually bring difficulty to the sample preparations for the
Prof. H. Wang, Prof. W. Chu conventional 3ω[13] or the time-domain thermoreflectance
CAS Key Laboratory of Nanosystem and Hierarchical Fabrication (TDTR)/frequency-domain thermoreflectance (FDTR) meas-
CAS Center of Excellence for Nanoscience urements.[14] Similar technical issues which are relevant to
National Center for Nanoscience and Technology
Beijing 100190, P. R. China
the characterizations of the OHT materials will be discussed
E-mail: wanghf@nanoctr.cn; wgchu@nanoctr.cn briefly in this paper. On the other hand, among these tech-
Prof. G. Chen niques some appear to be more critical to the understanding
College of Materials Science and Engineering of the thermal transport in the OHT materials. For example,
Shenzhen University one important feature associated with the OHT materials is
Shenzhen 518055, P. R. China the thermal conductivity anisotropy, which has become an
E-mail: chengm@szu.edu.cn
important research subject in the OHT community.[4,6] There-
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/aelm.201900167.
fore, this review will also put special emphasis on the tech-
niques capable of measuring the in-plane and cross-plane
DOI: 10.1002/aelm.201900167 thermal conductivity of the samples.

Adv. Electron. Mater. 2019, 1900167 1900167 (1 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

Based on the time dependence of the heating sources, the


techniques for charactering the thermal conductivity can be Hanfu Wang obtained his
categorized into two major groups, i.e., steady-state methods Ph.D. from the University of
and transient methods.[15] For the steady methods, a con- Minnesota at Minneapolis,
stant temperature gradient is created along the sample by USA in 2001 and pursued his
applying a time-independent heat flow. The thermal conduc- postdoctoral work at Pacific
tivity is then deduced with the knowledge of the amount of Northwest National Laboratory,
heat conducted through the sample, the resulted tempera- USA from 2001 to 2004. He
ture difference, and the sample dimensions. As for the tran- joined the National Center for
sient methods, a time-dependent heat flow is imposed on the Nanoscience and Technology,
sample and the corresponding thermal response is measured. China as an associate pro-
The transient methods can be further divided into two sub- fessor in 2004. His major
categories: time-domain methods and frequency-domain research interest is thermoelec-
methods. In the time-domain methods, the thermal response tric materials and devices.
is recorded and analyzed as a function of time right after the
sample being perturbed by a heating pulse. Unlike the time-
Weiguo Chu obtained his
domain method, the frequency-domain methods usually
Ph.D. in 2000 from Harbin
exert periodic thermal perturbation which in turn produces a
Institute of Technology. In
periodic signal whose phase and amplitude will be analyzed.
2004, he joined the National
Some transient methods (such as laser flash method) do not
Center for Nanoscience and
directly measure the thermal conductivity due to the difficul-
Technology, China. Currently,
ties of accurately measuring the heat flow conducted through
his research interests are
the sample. In these cases, it is often convenient to measure
quite broad, including energy-
the thermal diffusivity α or thermal effusivity e instead. The
related nanomaterials and
thermal diffusivity of a material measures the rate of the heat
devices, optical metasurfaces,
transfer in response to a temperature change, while the effu-
nanosensors, and micro/
sivity characterizes its ability to exchange thermal energy with
nanofabrication technologies.
its surroundings.[15] The relation of the three transport proper-
ties is expressed as
Guangming Chen obtained
κ = C Vα (2)
his Ph.D. in 2000 from
e = κ CV (3) Institute of Chemistry,
Chinese Academy of Sciences
(ICCAS) and pursued his
κ =e α (4)
postdoctoral work at National
Institute for Materials
where CV is volumetric heat capacity (J m−3 K−1), which can Science, Japan (NIMS) from
be determined from density ρ (kg m−3) and specific heat Cp 2000 to 2002. He worked
(J kg−1 K−1) through CV = ρCp. Therefore, κ can be determined at ICCAS between 2002
if any two quantities among α, e, and CV are known. and 2018. He then joined
The measurement techniques for the thermal conduc- Shenzhen University as a
tivity can also be divided into contact methods and noncon- distinguished professor. His current research interest is
tact methods, according to whether the sample is directly thermoelectric composites.
contacted with the heat source and the temperature sen-
sors. The noncontact methods generally involve with optical
heating and optical sensing. An advantage of the noncon-
tact methods is that the thermal contact resistance between 2. Steady-State Methods
the sample and the heat source/temperature sensors is
eliminated, which makes the high-accuracy measurement This section will cover six measurement techniques, namely,
possible. the absolute method, the parallel thermal conductance method,
In this work, we review commonly used and newly devel- the comparative method, the Völklein method, the method
oped techniques for the thermal conductivity measurement. based on microfabricated suspended devices, and the steady-
Most of them have been applied to the OHT materials in state infrared (IR) thermography. The bulk samples can be
recent years. The review is divided into three main sections measured by the former three techniques, while the in-plane
focused on steady-state, time-domain, and frequency-domain thermal conductivity of the thin films by the latter three ones.
methods. The operating principles, merits and limitations, Among these methods, the infrared thermography is a noncon-
technical issues, and application examples related spe- tact absolute method, while all others are usually configured as
cifically to the measurement of the OHT materials will be the contact methods. The advantages and limitations of these
discussed. techniques are summarized in Table 1.

Adv. Electron. Mater. 2019, 1900167 1900167 (2 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

Table 1. Summary of the steady-states methods presented in this review for measuring the thermal transport properties of the OHT materials.

Methods Measured quantities Merits Limitations


Absolute method Thermal conductivity of bulk materials Easy implementation; Can obtain κ; Possible to Heat losses and thermal contact resistances need
be measured with S and σ to be considered
Parallel thermal Thermal conductivity of bulk materials Can obtain κ; Easy implementation The thermal conductance of the sample holder
­conductance method needs to be small; Issue of thermal contact
resistance
Comparative method Thermal conductivity of bulk materials Can obtain κ; No need for measuring heat flow; Heat losses and thermal contact resistances need
Possible to be measured with S and σ to be considered
Völklein method In-plane thermal conductivity of films Substrate effect eliminated; Prefabricated metal The thermal conductance of the bare membrane
stripes and insulation layers; Compatible with needs to be small when compared to that of the
various film deposition techniques; Can obtain κ sample
directly; Possible to be measured with S and σ
Microfabricated Thermal conductivity of nanowires or Capable of characterizing the nanoscale or Extensive microfabrication works; Difficulty in
­suspended device nanotubes along their axial direction, microscale samples; Substrate effect eliminated; transferring as-prepared sample to the device;
or thin film stripes along their in-plane Can obtain κ directly; Possible to be measured Issues of thermal contact resistance and measure-
direction with S and σ ment sensitivity need to be considered
Steady-state infrared Thermal conductivity of films along the Rapid and noncontact; Can obtain κ directly; Needs sample with substantial optical absorptance
thermography in-plane direction Applicable to sub-micrometer thick films and large emissivity; Extensive calibrations of the
experimental setup

2.1. Absolute Method

Figure 1 illustrates a typical experimental configuration of the


absolute method for measuring the thermal conductivity of
bulk materials in low temperature region. The bar-like or rod-
like sample is attached with a heater at one end, while another
end of the sample is anchored to a heat sink. Two temperature
sensors, which could be thermocouples or resistance thermom-
eters, are placed along the sample to measure the temperature
difference ΔT. The sample is heated by a constant heating
power. When the temperature distribution along the sample
reaches the steady state, ΔT is recorded and the thermal con-
ductivity κ is calculated by

∆T
qs = κ A (5)
L

where qs is the heat flow through the sample, A is the cross-


sectional area of the sample. L is the distance between the two
temperature sensing locations.
Since heat exchanges between the sample and surround-
ings can happen through radiation, gas convection, and con- Figure 1. A typical experimental configuration of the absolute method
for measuring the thermal conductivity of bulk materials in the low-­
duction (e.g., through lead wires), the heat power dissipated temperature region.
from the heater may significantly differ from qs. Proper steps
should be taken in order to minimize the parasitic heat losses.
For example, a thermal shielding is often installed around the above 200 K).[17–19] Therefore, careful calibrations of the
sample to reduce the radiation exchange. To minimize the gas apparatus or proper corrections of the radiation heat losses
convection, the test setup is often installed in a vacuum envi- are essential for reliable measurements. With the configura-
ronment. To reduce the heat losses due to conduction, it is tion shown in Figure 1, Kovalsky et al. have measured the
preferable to use long and thin lead wires made of low thermal thermal conductivity of an organic–inorganic hybrid perov-
conductivity materials for the heaters and temperature sensors. skite CH3NH3PbI3 from 7 to 300 K.[20] The sample was in the
The thermal contact resistances between the sample and the form of pellet mechanically pressed from the CH3NH3PbI3
heater/sink as well as between the sample and the temperature powders. The data at temperatures above 50 K are in rea-
sensors should be minimized.[16] sonable agreement with those of a similar polycrystalline
It has been found that the radiation heat losses may cause sample measured by Pisoni et al. with the steady-state com-
substantial measurement errors at higher temperatures (e.g., parative method that will be discussed in Section 2.3.[21] In

Adv. Electron. Mater. 2019, 1900167 1900167 (3 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

The thermal resistance Rs of the sample is


given by

Th − Tc
Rs = (6)
qs

In principle, several samples with different


thickness d should be tested in order to elim-
inate the thermal contact resistance Rcontact
between the sample and the metal bars.[23,25]
The thermal conductivity of the sample can
be derived from the slope of Rs versus d
curve as

d
Rs = + Rcontact (7)

Figure 2. Two configurations for thermal conductivity measurement according to the ASTM In recent years, the methods based on the
D5470 standard. ASTM D5470 standard have been used to
measure the thermal conductivity of several
both studies, the thermal conductivity at room temperature is thermoelectric carbon nanotube (CNT)/polymer nanocompos-
around 0.3 W m−1 K−1. ites.[24,26–29] The thickness of the specimens ranged typically
Efforts have been made to carry out the thermal conductivity from several tens of micrometers to hundreds of micrometers.
measurement above room temperature with the same experi- It has been found that the polymer composites with randomly
mental configuration. Zaitsev et al. have demonstrated an appa- distributed CNTs demonstrate fairly low thermal conductivities.
ratus which employed a radiation shield thermally anchored For example, Yu et al. prepared a series of samples by mixing
to both the heater and the heat sink to simulate a t­emperature single-wall carbon nanotube (SWCNT) with poly(3,4-ethylen
gradient that existed in the sample.[22] To further reduce the edioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) and/
radiation heat loss, the space between sample and the heat or polyvinyl acetate polymers. The thermal conductivities for
shield was filled with thermally insulating powder. With this the composites with 60 wt% CNT were found to be 0.2–0.4 W
setup, the authors were able to simultaneously determine the m−1 K−1 at room temperature.[28] The values are close to those
thermal conductivity, electrical conductivity, and Seebeck coef- reported for multiwalled carbon nanotube (MWCNT)/polyani-
ficient of Mg2Si1−xSnx solid solution in a temperature range of line (PANI) composites, which lie between 0.4–0.5 W m−1 K−1
300–870 K.[22] when the MWCNT content was varied from 20 to 80 wt%.[27]
Another absolute steady-state method relies on test stacks In general, the absolute method is relatively easy to imple-
constructed from metal bars. The related test apparatus ment. By properly setting up the experimental apparatus, it is
and test procedures have been specialized in ASTM D5470 possible to measure the thermal conductivity along with the
standard.[23] Figure 2 shows two test configurations based on electrical conductivity and the seebeck coefficient for the bulk
this method. In the first configuration, the sample is pressed samples.[22] On the other hand, the heat losses and the thermal
between two metal bars (also known as meter bars) which are contact resistance are two major concerns when carrying out
made from highly thermally conductive materials with well- the absolute method. For the measurement of the OHT mate-
documented thermal conductivity (Figure 2a). A linear array rials, how to effectively reduce the thermal contact resistance
of temperature sensors is placed in each metal bar. When the between the sample surface and testing components (e.g.,
steady-state heat flow is reached, the heat flow through the meter bars, heater, heat sink, temperature sensors, etc.) is
sample can be measured by the metal bars. The interfacial especially critical, since the polymer or other organic mate-
temperatures between the sample and the metal bar (Th and Tc) rials usually possess relatively large thermal expansion coef-
are extrapolated from the linear temperature distributions on ficient which may be as high as 1.2 × 10−4 K−1.[9] To elucidate
both hot side and cold side.[24] In the second ­configuration, the problems caused by the samples with the large thermal
a guard heater is installed around the main heater in order expansion coefficient, it can be assumed that the sample is
to significantly reduce the heat leakage from the latter one[25] measured by the ASTM D5470 apparatus and the temperature
(Figure 2b). In this case, qs can be deduced from the power difference across the sample is large (e.g., >30 K). Under such
dissipated from the heater with necessary corrections for the a scenario, the thermal expansion at the hot surface of the
heat losses.[25] Th and Tc are obtained from the temperature sample may be significantly larger than that at the cold sur-
sensors that are installed in extreme proximity to the inter- face. This could cause the sample to bow and to generate gaps
faces between the sample and the metal bars. In both con- at the interfaces between the sample and the metal bars.[10]
figurations, compressive load and thermal interface materials The formation of the gaps leads to an increase of the thermal
such as thermal grease could be applied to minimize the contact resistance which may eventually deteriorate the quality
thermal contact resistances between the sample and the metal of the measurement . Therefore, to measure the OHT mate-
bars.[23] rials with relatively large thermal expansion c­oefficients, the

Adv. Electron. Mater. 2019, 1900167 1900167 (4 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

temperature difference should be carefully kept at a level (e.g., where L and A are the length and the cross-sectional area of the
several Kelvins[24]) that is small enough to avoid the formation sample, respectively.
of the gaps but is still large enough for carrying out reliable The technique is capable of obtaining the thermal conduc-
measurements. tivity of the small bulk samples with a relatively simple experi-
mental setup. It does not require special sample treatments
2.2. Parallel Thermal Conductance Method and the sample mounting is simple and rapid. Recently, Wan
et al. have applied this method to obtain the in-plane thermal
The parallel conductance method was developed by Tritt and conductivities of TiS2 single crystal and an inorganic/organic
co-workers for measuring small bulk samples at the millimeter superlattice material TiS2[(HA)0.08(H2O)0.22(DMSO)0.03]
scale which often bring about difficulties to the temperature at room temperature, which were found to be 4.45 and
and heat flow measurements.[30] As shown in Figure 3a, the 0.69 W m−1 K−1, respectively.[31] The value of the TiS2 single
sample holder consists of a heater (hot end) and a heat sink crystal is close in magnitude to the result previously reported
(cold end), which are connected by a supporting post with low by Imai et al.[32] Varghese et al. have measured the in-plane
thermal conductance. A differential thermocouple is used to thermal conductivity of an ≈100 µm thick free-standing ther-
measure the temperature difference between the hot end and moelectric film made by screen printing colloidal inks com-
the cold end. The measurement is carried out in two steps. In posed of Bi2Te2.8Se0.2 nanocrystals. The value (≈0.56 W m−1 K−1
the first step, a baseline thermal conductance Gbaseline is meas- at room temperature) is within 5% of the cross-plane thermal
ured for the empty sample holder, which accounts for the heat conductivity determined by the laser flash method, which indi-
losses associated with the sample holder itself. To do this, mul- cates the isotropic nature of the sample.[33]
tiple DC heating currents (I’s) are applied to the heater suc- Two issues should be paid attention when implementing the
cessively to generate corresponding steady-state temperature technique. First of all, to maintain a reasonably high measure-
differences (ΔT’s) across the supporting post. The heating ment sensitivity, the baseline thermal conductance associated
power P varies linearly with ΔT as denoted by the following with the sample holder Gbaseline needs to be made as low as
equation possible.[30] Second, the thermal contact resistance between the
specimen and the sample holder should be kept much smaller
P = I 2Re,H = Gbaseline ∆T (8) than the thermal resistance of the specimen itself to ensure the
accuracy of the measurement.
where Re,H is the resistance of the heater. Gbaseline is extracted
from the slope of the P ~ ΔT plot. The sample is then attached
to the sample holder, and the total thermal conductance Gtotal is 2.3. Comparative Method
measured in a similar way
P = I 2Re,H = Gtotal ∆T (9) The comparative method determines the thermal conductivity
of bulk materials without measuring the heat flow through
Both Gbaseline and Gtotal are measured in a high vacuum envi- the sample qs. In this method, a reference sample with known
ronment to reduce the convective heat losses.[31] The thermal thermal conductivity is thermally connected in series with the
conductivity κ of the sample is determined from the following specimen to be investigated. The stack structure is sandwiched
equation between the heater and the heat sink. Two or more temperature
κ = (Gtotal − Gbaseline ) L/A (10) sensors are arranged along the test sample and the reference
sample, respectively (Figure 4). Under the steady-state condi-
tion, the heat flow through the sample should be equal to that

Figure 3. Schematic diagram of the parallel thermal conductance method. Figure 4. Schematic of the experimental setup for the comparative
a) Empty sample holder. b) Sample holder attached with a sample. method.

Adv. Electron. Mater. 2019, 1900167 1900167 (5 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

through the reference sample. The thermal conductivity of the considered to be a merit for this modification. However, sig-
sample is then given by nificant error in measuring κ has been found in their studies
presumably due to high residual gas pressure remained in
 A ∆T L  their measurement chamber.[40] This implies that the issues
κ s = κ ref  ref ref s  (11)
 As ∆TsLref  such as heat losses and thermal contact resistance require
extremely careful handling when characterizing the small
where subscripts “s” and “ref” denote the specimen to be tested samples.[38]
and the reference sample, respectively. The method achieves
the best accuracy if the thermal conductivity of the sample is
comparable with that of the reference sample.[34] 2.4. Völklein Method
Test standards like ASTM E1225[35] and ASTM E1530[36]
describe the relevant apparatuses and test procedures for the Völklein et al. have developed a chip-based method to measure
comparative method. When applying this method to the OHT the in-plane thermal conductivity of the sub-micrometer thick
materials, the influence of the uneven thermal expansions thin film.[42–46] The chip (known as λ-chip) includes a rectan-
on the thermal contact resistance should be minimized[10] as gular suspended dielectric (e.g., SiNx) membrane with a thick-
for the absolute method discussed in Section 2.1. Marconnet ness of 50 to 100 nm, on which the film specimen can be
et al. have characterized the thermal conductivity of aligned deposited (Figure 5a). A thin metal stripe is created along the
CNT-thermoset epoxy composites using a setup conforming long axis of the membrane which serves as both a heater and
to the ASTM E1225 standard.[37] In this study, the tempera- thermometer. The silicon rim surrounding the membrane acts
ture sensing was carried out with an IR microscope instead as a heat sink. All measurements should be conducted in the
of thermocouples. The axial thermal conductivity of the com- vacuum environment. The thermal conductivity of the bare
posites was found to increase from 0.46 to 4.87 W m−1 K−1 as membrane needs to be determined first. To do this, a dc cur-
the volume fraction of CNT varied from 1 to 16.7 vol%, which rent is applied to the metal stripe and the resulted heat will
is comparable with that of unfilled and aligned CNT arrays flow through the membrane to the heat sink. At the same time,
investigated in the same study. On the other hand, as men- radiation heat loss will occur from the surfaces of the mem-
tioned in Section 2.1, Pisoni et al. have measured the tem- brane and the metal stripe. From the temperature rise and
perature-dependent thermal conductivity of the CH3NH3PbI3 the input electrical power, an apparent thermal conductance
sample[21] by adopting the experimental configuration shown in associated with the chip (G) can be calculated. With certain
Figure 4 and the obtained results were confirmed later by other approximations, it can be shown that the thermal conductance
authors.[20] of the membrane due to heat conduction and radiation (GM)
Verebelyi has extended this method to measure the is roughly equal to G.[44] Since GM is a function of unknown
thermal conductivity of Pb-doped Bi2Sr2Ca1Cu2Ox whisker thermal conductivity κM,∥ and emissivity ε of the membrane,
with d­ imensions less than 1000 × 100 × 100 µm3 within a it is necessary to carry out two measurements with two dif-
­temperature range of 30–200 K.[38] The measurement error ferent membrane geometries so that one may obtain κM,∥ and ε
for such a small sample was estimated to be less than 25% simultaneously.[44]
with the application of the temperature matching shielding The thermoelectric thin film to be tested can be deposited
and micro-thermocouples. Later, Yoshino et al. have applied on the top side (top configuration)[46] or bottom side (bottom
this method to measure the thermal conductivity of mil- configuration)[45] of the membrane through various vapor dep-
limeter-sized organic crystals. By properly designing the osition procedures such as thermal evaporation, magnetron
sample holder, they were able to simultaneously obtain the sputtering, atomic layer deposition, etc. If the film is depos-
values of the thermal conductivity, the Seebeck coefficient, ited on the top side of the membrane, the metal stripe should
and the electrical conductivity from the same sample.[39–41] be covered with a thin insulating layer in order to avoid elec-
The capability of measuring κ together with S and σ is trical shortcuts (Figure 5b). The top configuration allows one to

Figure 5. a) Schematic illustration of the measurement chip used in the Völklein method. b) Deposition of the thermoelectric film on top of the
­suspended membrane.

Adv. Electron. Mater. 2019, 1900167 1900167 (6 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

prepare the OHT samples by different kinds of solution-based 2.5. Microfabricated Suspended Device
methods such as spin-coating, dip-coating, drop casting, ink-jet
printing, etc.[47,48] Aiming for eliminating the parasitic heat transfer from the
To obtain the in-plane thermal conductivity of the depos- sample into the substrate, another steady-state method based
ited film (κf,∥), the thermal conductivity of the composite on microfabricated suspended devices (MFSD) has been devel-
layer (including the test film and the supporting mem- oped to measure the thermal conductivity of nanowires or
brane) κC,∥ is measured by employing the same procedure nanotubes along their axial direction or narrow thin film stripes
of obtaining κM,∥. κf,∥ is subsequently extracted from the fol- along their in-plane direction.[50–54] The temperature sensing of
lowing equation[44,45] this technique is performed at both ends of the sample and all
measurements are carried out in the vacuum environment to
reduce the convective heat loss. Two typical configurations will
κ C, dC − κ M, dM
κ f , = (12) be reviewed in this paper.
df One configuration proposed by Shi et al.[50] is schematically
shown in Figure 6a in which the test platform consists of two
where df, dM, dC are the thicknesses of the thin film, the mem- adjacent suspended SiNx membranes each supported by five
brane, and the composite layer, respectively. By applying this identical long and thin SiNx beams. A platinum resistance ther-
method, Linceis et al. have recently measured κf,∥ for a 110 nm mometer (PRT) is fabricated on each membrane to serve as the
thick Bi87Sb13 film over a temperature range of 120–450 K,[45] heater or the thermometer. The sample is transferred onto the
which shows good agreement with previously published device to bridge the two membranes. One membrane is used as
results.[49] heating membrane on which the PRT is heated by a dc current,
The use of the ultrathin suspended membrane is a major while another one is used as sensing membrane. It is assumed
advantage of this method, since the parasitic heat transfer from that the temperature rise on both membranes is uniform and
the sample into the substrate is eliminated,[43] which is critical to the heat losses due to gas flow, convection, and radiation can
achieving reliable measurements on the sub-micrometer thick be neglected. The Joule heat generated by the heater and the
thermoelectric films.[44,45] Additionally, the insulation layer two Pt current leads on the heating membrane are PH and 2PL,
and metal stripes are prefabricated on the test chip, which not respectively. A certain amount of the generated heat is trans-
only significantly simplifies the sample preparations but also ferred from the heating membrane directly to the environment
prevents the samples from undergoing harsh environments through the five beams. The rest heat flows into the sensing
of depositing the insulation layer or performing lithography membrane through the sample and is further conducted to
process. This is very important for the characterization of the the environment through the five beams connected with the
vulnerable OHT materials.[13] Moreover, the current experi- sensing membrane. It can be shown that the measured thermal
mental configuration has been integrated with measurement conductance of the sample Gs is expressed as[50]
techniques of the other TE properties on a single chip, there- ∆Tc (PH + PL )
fore all quantities needed for calculating the in-plane ZT-value Gs = (13)
∆Th2 − ∆Tc2
can be obtained from the same sample in one measurement
run.[45] where ΔTh and ΔTc are the temperature rise at the heating
On the other hand, it should be noted that the method is membrane and the sensing membrane, respectively. Both
incapable of measuring the cross-plane thermal conductivity of ΔTh and ΔTc can be measured from the resistance change of
the thin films. The measurement must be carried out in the the PRT. On the other hand, the measured thermal resistance
vacuum environment. The measurement accuracy depends on (i.e., Gs−1) consists of the intrinsic thermal resistance of the sample
the relative magnitudes of the thermal conductance associated Rn, and the thermal contact resistance between the sample and
with the deposited film and the bare membrane. If possible, the the two membranes, Rcontact
thermal conductance of the bare membrane should be made
smaller than that of the deposited film (κM,∥dM < κf,∥df). Gs−1 = R n + Rcontact (14)

Figure 6. a) Schematic illustration of the experimental configuration proposed by Shi et al.[50] b) Another configuration proposed by Sultan et al.[53]

Adv. Electron. Mater. 2019, 1900167 1900167 (7 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

Only if Rcontact is small enough, the thermal conductivity of the including oxidized polyacetylene nanofibers (74–126 nm
sample κ can be calculated directly from Gs−1 by in diameter),[59] amorphous polythiophene nanofibers
L L (71–245 nm in diameter),[60] PEDOT:PSS and PEDOT:Tosylate
κ = s ≈ −1s (15) (Tos) films (130–230 nm thick),[61] and polymer treated CNT
Rn As Gs As
films or polymer:CNT hybrid films (≈90 nm thick in ref. [62]
where Ls and As are the length and cross-sectional area of the and 80–110 nm thick in ref. [64]).[62–64] Among these studies,
sample, respectively. Therefore, how to reduce Rcontact is one of Weathers et al. have systematically investigated the in-
major concerns when applying this technique. In practice, Rcontact plane thermal conductivities of a series of PEDOT:PSS and
can be reduced by metal or amorphous carbon deposition at PEDOT:Tos films, which demonstrated an obvious correlation
the locations where the sample is contacted with the mem- with the in-plane electrical conductivity σ∥. As σ∥ changed from
branes.[50] Alternatively, new techniques have been invented 20 to 500 S cm−1, the thermal conductivity could vary from
later to explicitly obtain the thermal contact resistance and about 0.5 W m−1 K−1 to as high as 1.8 W m−1 K−1 at 300 K.[61]
the intrinsic thermal resistance of the sample by, for example, The value of 0.5 W m−1 K−1 is close to the in-plane thermal con-
making use of the thermoelectric voltage of the sample to deter- ductivity (≈0.58 W m−1 K−1) for a low-σ∥ PEDOT:Tos film meas-
mine the temperature drops at the contacts.[51,55] ured by Ushirokita and Tada using a chip-based 3ω method that
Figure 6b illustrates another experimental configuration pro- will be discussed in Section 4.3.[65]
posed by Sultan et al. in which two identical suspended SiNx It is also worth mentioning that the test platforms can be
membranes each equipped with the heater and thermometer designed to couple with the characterizations of the Seebeck
and supported by four legs are linked together by a narrow SiNx coefficient and the electrical conductivity of the same sample.[61]
bridge.[53,54,56] One membrane is used as the heating mem- On the other hand, the SiNx bridge in Figure 6b allows direct
brane excited by a dc current, while another one as the sensing depositions of the films on it by various physical vapor deposi-
membrane. It can be demonstrated that the steady-state tem- tion (PVD) techniques[53,54] as well as solution-based ultrasonic
perature rises on the two membranes are related to the heating spray (UP) process.[62,63] The combination of the MFSD with
power P by[53] the UP process will facilitate the measurements of those solu-
tion- processable OHT films.[62,63]
GL + GB The major drawbacks associated with the method are that
∆Th = P (16)
(2GB + GL )GL it requires extensive microfabrication works and the transfer
of as-prepared samples to the microdevices could be chal-
GB lenging. The thermal contact resistance and the measure-
∆Tc = P (17)
( 2GB + GL )GL ment sensitivity are also critical issues that needed to be
carefully considered. Since the measurement needs to be car-
where GB and GL are the thermal conductances of the bridge ried out under the high vacuum condition, the technique is
and the supporting legs, respectively. GB and GL can be solved not applicable to the humidity-controlled tests for the OHT
from the slopes of the ΔTh ∼ P and ΔTc ∼ P straight lines. Before materials.[66]
transferring the test sample onto the bridge, the thermal con-
ductance of the bare bridge GB,Si-N needs to be measured first.
Then the thermal conductance GB,C of the composite structure 2.6. Steady-State Infrared Thermography
consisting of the sample and the SiNx bridge is measured in the
same manner. The difference between GB,C and GB,Si-N is used The steady-state infrared thermography has been employed
to compute the thermal conductivity of the sample. It should to determine the in-pane thermal conductivity of the free-
be noticed that thermal contact resistance between the sample standing PEDOT:PSS film[67] and PEDOT:PSS/Si nanoparticle
and the two membranes should also be kept as low as possible. composite films.[68] As shown in Figure 7, the thin film sample
Besides the thermal contact resistance, another impor- is suspended over a cavity in an opaque substrate with high
tant issue associated with the MFSD method is the measure- thermal conductivity. The sample is homogenously illuminated
ment sensitivity of the device, which sets the lower limit of the from below by visible light with an absorbed power density QA
thermal conductance that can be measured for a sample.[8] This
value is around 1 nW K−1 for a typical device with the configu-
ration shown in Figure 6a.[50] The measurement sensitivity is
limited by several factors, including the heat transfer through
residual gas molecules, the radiation from the membranes,
and temperature fluctuation of the sample environment.[57]
Recently, significant progresses have been made to increase
the measurement sensitivity, it is now possible to measure the
sample with a low thermal conductance at the order of 10−10
or 10−11 W K−1.[57,58] For the configuration shown in Figure 6b,
the major factor that affects the measurement sensitivity is the
background thermal conductance of the bridge.
Both configurations have been used to measure the thermal
conductivity of polymer or OHT materials in recent years, Figure 7. Schematic of steady-state infrared thermography.

Adv. Electron. Mater. 2019, 1900167 1900167 (8 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

in a vacuum chamber. The wavelength of the light is carefully thermal diffusivity and the thermal conductivity of the bulk
selected to cause strong optical absorptance for generating sub- samples, ­respectively. The time-domain thermoreflectance
stantial heating. The substrate serves both as a shadow mask method, on the other hand, is suitable for characterizing the
for the illumination and as a heat sink for the heat flow. The samples ranging from the bulks to sub-micrometer thick films.
temperature distribution is recorded by an infrared camera. Table 2 summarizes the major merits and limitations of these
Theoretically, the temperature at a distance r from the cavity three methods.
center is given by[67]

QA QA 3.1. Laser Flash Method


T (r ) = − r2 + r02 + T0 (18)
4κ f , df 4κ f , df
The laser flash method was first introduced by Parker et al. in
where r0 is the radius of the cavity. κf,∥ and df are the in- 1961 and has gained widespread acceptance since then.[71] It
plane thermal conductivity and the thickness of the film, is a noncontact technique for characterizing the thermal dif-
­respectively. T0 is the temperature of the heat sink. Since the fusivity of the solid materials.[7,72,73] In this method, the front
last two terms in Equation (18) are just constants, κf,∥ can surface of the sample is irritated uniformly by a short laser
be deduced from the parabolic temperature distribution near pulse, the instantaneous temperature rise on the rear side
the center of the cavity, the absorbed power density, and the sample is recorded by a fast detector (e.g., infrared detector)
film thickness. as a function of time. The use of the infrared thermographic
The steady-state infrared thermography represents a non- technique avoids the thermal contact resistance that will exist
contact absolute method which can directly yield the in-plane between the sample and a conventional thermometer such as
thermal conductivity of the sub-micrometer thick thin films. the thermocouple.
However, extensive calibrations of the light source and the IR A very thin layer of graphite or other high emissivity coating
camera are required for performing the measurements. In is usually applied on both sides of the sample to enhance
order to apply the method to the OHT materials, it is neces- the light absorption on the front surface of the sample and
sary for the samples to have a substantial absorptance at the infrared emission on the rear side of the sample.[74] If the
illuminating wavelength and a relatively large emissivity for heat conduction from the front surface to the rear surface
the IR detection.[67] In ref. [67], the 250 nm thick PEDOT:PSS is assumed to be 1D and heat losses from the sample are
film (σ∥ = 720 S cm−1) post-treated by ethylene glycol (EG) neglected, the thermal diffusivity of a homogenous sample
was found to have an in-plane thermal conductivity of can be expressed as[7]
1.00 W m−1 K−1 at room temperature, which is comparable
with the literature data for highly electrically conductive sam-  d2 
α = 1.37  2
(19)
ples such as dimethyl sulfoxide (DMSO)-mixed PEDOT:PSS  t1/2π 
film (κf,∥ is estimated to be 1.05 W m−1 K−1 at σ∥ ≈ 520 S cm−1)
measured by the TDTR method[69] and EG-mixed PEDOT:PSS where d is the thickness of the sample. t1/2 is the time needed
sample (κf,∥ = 0.84 W m−1 K−1 at σ∥ = 820 S cm−1) character- for the temperature on the rear surface (T(t)) to reach half of
ized by the laser flash method.[70] its maximum temperature rise (Tmax ), which is schematically
illustrated in Figure 8.
The major advantage of the method is that it allows the
3. Time-Domain Methods thermal conductivity being measured rapidly and reliably over a
wide temperature range (148–3073 K),[15] provided that the den-
This section reviews three time-domain methods, including sity and the specific heat of the sample have been obtained sep-
two noncontact methods (i.e., the laser flash method and the arately. On the other hand, the method requires surface coating
time-domain thermoreflectance method) and one contact which prevents the sample from being used for other test pur-
method (i.e., the hot-disk method). The laser flash method poses. Also, it is not capable of measuring the sub-micrometer
and the hot disk method are commonly used to measure the thick thin films as the thickness for the sample that can be

Table 2. Summary of the time-domain methods presented in this review for measuring the thermal transport properties of the OHT materials.

Methods Measured quantities Merits Limitations


Laser flash method Cross-plane thermal diffusivity of bulk Rapid and noncontact; Needs Cp and ρ;
materials mainly wide temperature range Thickness limitation for the conventional laser flash
method; Surface coating needed
Time-domain Thermal conductivity in the cross-plane Rapid and noncontact; Complicated experimental setup; Sophisticated data
thermoreflectance direction mainly, possible in the in-plane May obtain κ without knowing CV; Applicable reduction procedure;
(TDTR) method direction to samples ranging from sub-micrometer Large uncertainty for κ∥ when κ∥ < 5 W m−1 K−1; Smooth
thick films to bulks sample surface needed; Needs surface coating
Hot-disk method Cross-plane thermal diffusivity and Nondestructive and rapid; Inapplicable to sub-micrometer thick films;
thermal conductivity of bulk materials Can obtain α and κ simultaneously over a wide Questionable for the measurements of low density
temperature range materials

Adv. Electron. Mater. 2019, 1900167 1900167 (9 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

3.2. Time-Domain Thermoreflectance Method

The TDTR method is an optical pump–probe technique for


measuring the thermal conductivity and interfacial thermal
conductance of thin films and bulk materials.[74,82–84] In this
method, a pump laser pulse at a fixed wavelength irradiates
the sample surface to generate a transient change in tem-
perature, which consequently causes a change in surface
reflectance that can be detected by a probe laser. The sample
surface is usually coated with a thin metallic film (i.e., trans-
ducer layer) whose reflectance changes with the temperature
rise at the wavelength of the probe laser. The temporal evo-
lution the thermoreflectance signal is schematically displayed
in Figure 9a. After ­initially heated by the short laser pulse,
Figure 8. Schematic of the laser flash method. a sudden jump in surface temperature is followed by a slow
cooling process from which the thermal properties of the
sample can by extracted by comparing with suitable theoret-
measured is limited by the pulse width of the heating laser and ical models.[74,85,86]
the response speed of the detector. The commercial i­ nstruments Depending on the pulse width of the pump laser and the
generally require the thickness of the sample to be larger than time scale for data acquisition, the TDTR methods can be
100 µm.[74] Therefore, many OHT materials investigated by the performed in the nanosecond domain for characterizing
laser flash method are made in the form of bulk.[75–79] Ao et al. the thermal properties of thin films,[87–91] in the picosecond
have applied the method to measure the thermal conductivity domain for measuring the thermal properties of thin films
of Bi2Te3/polythiophene composites, which shows a weak tem- and their ­interfacial properties,[82,92] and in the femtosecond
perature dependence. For example, the thermal conductivity regime for investigating electron and phonon processes.[90,93,94]
of the composite with 30 wt% polythiophene was found to be If the thermoreflectance technique is applied in the nano-
around 0.33 W m−1 K−1 in a temperature range of 300–473 K.[76] second domain, a continuous wave (CW) laser can be used as
A similar observation has been made for Sb2Te3/PEDOT com- the probe laser.[87–91] On the other hand, when working in the
posite, which demonstrates a nearly constant value around 0.15 picosecond or femtosecond domain, an ultrafast pulsed laser
W m−1 K−1 between 300 and 523 K.[78] Yao et al. have shown that system should be used as the light source for fulfilling the
the thermal conductivity of the SWCNT/PANI composite varies pumping and probing purposes.[74,84,95] Figure 9b schemati-
from ≈0.34 to 0. 48 W m−1 K−1 as the SWCNT content increases cally depicts such an experimental setup in which an ultrafast
from 30 to 89 wt.%,[80] which is in reasonable agreement with pulsed laser light is split into the pump beam and the probe
the results previously obtained by the absolute steady method beam by a polarizing beam splitter (PBS). By varying the time
for the MWCNT/PANI composites.[27] delay between the probe pulse and the pump pulse using a
To extend the measurement capability of the conventional mechanical delay stage, the information of the temporal evolu-
laser flash method into the thin film region, a new technique tion of the surface temperature can be recorded. In order to
known as ultrafast laser flash method has been developed by increase the signal to noise ratio, the train of the pump pulses
using the picosecond laser as the light source and the thermore- is modulated at frequency f mod by an electric-optic modulator
flectance as the temperature sensing mechanism.[81] Unlike the (EOM) so that the thermal response signals can be detected
TDTR or FDTR methods which use the front heating/front by a lock-in amplifier.[96] The modulation frequency f mod typ-
detection configuration (please refer to Sections 3.2 and 4.7), ically ranges from 1 MHz to around 10 MHz, which makes
the ultrafast laser flash method uses the rear heating/front it possible to constrain the thermal penetration depth within
detection configuration which can be viewed as a natural exten- the sub-micrometer range.[84,86] Hence, the TDTR method
sion of the conventional laser flash method.[81] has become an attractive tool in characterizing the thermal

Figure 9. a) Schematic of the temporal evolution of the thermoreflectance signal. b) Sketch of the TDTR experimental setup based on an
­ultrafast-pulsed-laser system.

Adv. Electron. Mater. 2019, 1900167 1900167 (10 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

conductivity of the sub-micrometer thick thin films, since the could reach 1.8 W m−1 K−1 when σ∥ is around 400–500 S
influences from the substrate can be eliminated.[97] Depending cm−1. The difference might be attributed to the fact that the
on the implementation strategies of the method, the extraction PEDOT samples in both studies have been prepared in dif-
of the thermal conductivity from the measurement data may or ferent manners.
may not require that the volumetric capacity CV of the sample The major limitations of the TDTR method include its
be predetermined.[98–102] In fact, it is possible to measure κ complicated experimental setup and sophisticated data-
and CV simultaneously by using a frequency-dependent TDTR reduction procedure as well as the surface coating. More-
method.[101,102] Though the TDTR method was originally devel- over, to perform the TDTR measurement, the surface of
oped to measure the cross-plane thermal conductivity κ⊥, κ∥ the sample needs to be smooth in order to minimize diffuse
may be measured by orientating the sample in such a way reflections.[14] The requirement may bring challenge to the
that the incident laser beam is parallel to the in-plane direc- sample preparations for certain OHT materials whose sur-
tion of the sample.[69,103] Alternatively, the TDTR method can face roughness might be difficult to control. For example,
be extended to characterize κ∥ by adopting variable spot size undulations or striations can naturally occur on the surface
configuration or beam-offset configuration.[14,96,104] of spin-coated or casted polymer films.[13] The OHT materials
The TDTR method provides a rapid and noncontact way can also possess strong anisotropy in thermal conductivity.
to obtain the thermal conductivity for the materials ranging To properly extract the cross-plane thermal conductivity of
from sub-micrometer thick films to bulks with a high lateral the anisotropic materials, it has been suggested that a large
spatial resolution.[103] It is not involved with the complicated spot size of the laser and high modulation frequency should
sample preparation process as required by the conventional be used to eliminate radial heat transport.[96] The measure-
3ω method that will be discussed Section 4.1. The method ment of the in-plane thermal conductivity based on variable
has attracted much attention in recent years as an important spot size configuration or beam-offset configuration[14,96,104]
technique for characterizing the thermal conductivity of var- seems even more challenging, since more complicated experi-
ious OHT materials including PEDOT:PSS films (100–400 nm mental setup and data reduction models are involved. Jiang
thick in ref. [99], 2.5 µm thick in ref. [100], and 80 nm thick et al. have suggested that it is difficult to measure the in-
in ref. [105]),[69,99,100,105] poly(3-hexylthiophene-2,5-diyl) (P3HT) plane thermal conductivity that is less than 5 W m−1 K−1 by
thin films (77–200 nm thick),[98] graphene based organic nano- using the configurations mentioned above due to limits in
composite (65 nm thick),[106] and tetracyanoquinodimethane the lowest modulation frequency and the smallest spot size
incorporated Cu3(BTC)2 (BTC = benzene tricarboxylate) metal– of the laser used in the current TDTR setups.[14] Although
organic framework films (200 nm thick).[101] With this method, it is technically possible to further reduce the low limit of
Liu et al. have observed that the cross-plane thermal conduc- the modulation frequency and the laser spot size in the
tivity κ⊥ of the drop-cast DMSO-mixed PEDOT:PSS films future,[14] the application of these configurations in meas-
does not change with the in-plane electrical conductivity σ∥ at uring the in-plane thermal conductivity of the OHT materials
≈325 K.[69] The average value of κ⊥ is about 0.3 W m−1 K−1, may be limited at this moment, since many OHT materials
which is consistent with the data obtained by the conven- of interest have an in-plane thermal conductivity less than
tional 3ω technique.[107] In the same study, the authors were 5 W m−1 K−1.[47,61,63,64,67]
able to obtain κ∥ for the PEDOT:PSS films by properly ori-
entating the sample embedded in an epoxy matrix with
respect to the direction of the laser beam.[69] The κ∥ value 3.3. Hot Disk Method
demonstrates an increasing trend with σ∥, which has also
been observed by Weathers et al. via the MFSD technique The hot disk method, also known as transient plane source
for the suspended PEDOT:Tos films prepared by vacuum (TPS) method,[108–111] uses a thin metal strip or a metal disk
vapor phase polymerization and chemical polymerization.[61] (usually with a bifilar spiral pattern as shown in Figure 10a)
The κ∥ value estimated from the published data in ref. [69] as a plane heat source which is sealed between two thin insu-
is ≈1.05 W m−1 K−1 for σ∥ ≈ 520 S cm−1, while κ∥ in ref. [61] lating polymer films. To perform the measurement, the encap-
sulated hot disk is sandwiched between two
identical samples and a constant electrical
current is applied to the heater (Figure 10b).
The temperature rise is monitored by meas-
uring the resistance change of the heater
as a function of time. To ensure that the
sample can be treated as thermally infinite-
size solid, a criterion has been put forward
as a guideline for preparing the sample with
proper dimensions, which states that the
distance from the sensor edge to the nearest
free boundary of the sample should be
larger than 2 α tmax , where α is the thermal
Figure 10. a) The bifilar spiral structure of a hot-disk sensor. b) Cross-sectional view of the diffusivity of the sample and tmax is the max-
experimental setup for the hot-disk method. imum transient recording time.[109,110,112]

Adv. Electron. Mater. 2019, 1900167 1900167 (11 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

Theoretically, the temperature increase detected by the bifilar It should be mentioned that Coquard et al. recently
spiral sensor is given by[109] s­uggested that the accuracy of the TPS method becomes
questionable when it is applied to low density materials
P
∆T (τ ) = D (τ ) (20) (such as polymer foams) due to the fact that their thermal
π 3/2rsenκ inertia is usually two orders of magnitude smaller than
that of the probe.[119] Therefore, if the method is applied to
αt aerogel-based or polymer foam-based OHT materials,[123,124] it
τ= (21)
rsen might be a good idea to crosscheck the results with another
method.[125,126]
where P is the output power of the heater; rsen is the radius
of the sensor; t is the time and τ is a dimensionless quantity
known as characteristic time ratio; D(τ) is a complicated dimen- 4. Frequency-Domain Methods
sionless time-function that can be evaluated numerically.[112]
An advantage of the hot disk method is that both thermal In the frequency-domain methods, the periodic thermal per-
conductivity κ and the thermal diffusivity α can be obtained turbations are often created by the resistive Joule heating or
from a single transient recording.[109,110] Hence, the volumetric laser heating. The thermal responses captured by various
specific heat can be obtained at the same time according to types of sensors (e.g., the resistive thermometer, thermo-
Equation (2). One way of extracting the thermal transport prop- couple, IR sensor, photodiode, acoustic sensor, etc.) are usu-
erties from the raw data is given as follows. α is varied as a trial ally fed into lock-in amplifiers to improve the signal-to-noise
parameter to yield a series of ΔT(τ) versus D(τ) plots. The cor- ratio. In this section, we will review eleven frequency-domain
rect value of α corresponds to the plot that ΔT(τ) varies linearly methods, including eight contact methods and three noncon-
with D(τ). From the slope of the ΔT(τ) versus D(τ) curve (i.e., tact methods. The contact methods include the traditional 3ω
P/π3/2rsenκ), one may immediately obtain the thermal conduc- method, the self-heating 3ω method, the 3ω method based
tivity κ.[112] To improve the accuracy of the result, Equation (20) on the suspended ultrathin dielectric membranes, the scan-
can be further modified by including a number of factors such ning thermal microscopy (SThM)-3ω method, the Ångström’s
as the heat capacity of the probe, the thermal resistance due method, the temperature wave analysis method, the light
to the insulation layer of the sensor, resistance change of the irradiated ac-calorimetric method, and the photo-thermoelec-
sensor with temperature, and so on.[109,112,113] An inconven- tric method, while the frequency-domain thermoreflectance
ience associated with the usual hot disk method is that the method, the modulated photothermal radiometry, and the
measurement requires two pieces of samples each with an photoacoustic method belong to the noncontact methods. The
identical planar surface. Recently, a new technique known as merits and disadvantages of these methods are summarized in
modified transient plane source method has been introduced Table 3.
to address this issue. The new method is based on one-sided
sensor, which allows one to carry out the test with one single
piece of sample.[114–118] 4.1. Traditional 3ω Method
The technique provides a nondestructive and rapid tool to
investigate the thermal conductivity and thermal diffusivity of The traditional 3ω method developed by Cahill and Pohl in
various materials over a wide temperature range.[119] Compared 1980s was initially used to measure the thermal conductivity
with the steady state methods, the size of the sample could be of bulk materials and thick films.[127,128] It was later extended
significantly small and the sample with irregular shape can be to measure the cross-plane thermal conductivity of the thin
accepted. During the test, the temperature rise caused by films on substrate.[129,130] The technique utilizes a thin metal
the transient heating is typically less 5 K,[111] which represents stripe (with a width ranging from several micrometers to
a small temperature perturbation imposed on the sample. This tens of micrometers) deposited on top of the specimen to
will benefit the measurement of the OHT materials which may serve as both a heater and thermometer. For electrically
suffer mechanical instability caused by the uneven thermal conductive specimens, a thin insulation layer is needed to
expansion under a large temperature difference as discussed separate the metal stripe from the sample (Figure 11). An
in Section 2.1. In recent several years, the technique has been ac current at frequency ω in the form of I(t) = I0cos(ωt) is
extensively used for characterizing the thermal conductivity of injected into the heater/sensor to create the Joule heating at
the OHT materials with a thickness ranging from several hun- frequency 2ω
dred micrometers to 1–2 millimeters.[115–118,120–123] Unfortunately,
the method is not applicable to measuring the thermal conduc- I 02Re,H
tivity of sub-micrometer thick films. With this technique, Wu
P(t ) =
2
(1 + cos(2ω t)) (22)
et al. measured the thermal conductivity of pellet samples (σ is
around 0.2–0.8 S cm−1) mechanically pressed from p-toluenesul- where I0 is the amplitude of the ac current and Re,H is the
fonic ­acid-doped PANI nanowires or nanorods.[117] The thermal resistance of the metal stripe. A temperature oscillation at
conductivity is about 0.3–0.4 W m−1 K−1 at 300 K, which is ­frequency 2ω (ΔT2ω) arises in response to the ac heating, which
in good agreement with that measured via the absolute in turn brings about a resistance fluctuation at the same fre-
steady-state method for a PANI sample (κ ≈ 0.38 W m−1 K−1 at quency. Multiplication of the 2ω resistance change by the ac
σ = 0.45 S cm−1) prepared in a similar fashion.[27] driving current gives rise to a small 3ω voltage component that

Adv. Electron. Mater. 2019, 1900167 1900167 (12 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

Table 3. Summary of the frequency-domain methods presented in this review for measuring the thermal transport properties of the OHT materials.

Methods Measured quantities Merits Limitations


Traditional 3ω method Thermal conductivity of bulk materials; Can obtain κ directly; Microfabrication of the metal stripe;
Thermal conductivity of thin films in Applicable to sub-micrometer thick thin films; Needs good insulation layer for the
the cross-plane direction mainly, also Insensitive to the radiation and convective heat losses electrically conductive samples
possible in the in-plane direction
Self-heating 3ω method Thermal conductivity of the fibers or Can obtain κ directly; Possible to be measured with Needs electrically conductive sample
nanowires along their axial direction, S and σ with a substantial TCR; The resistance
or the thin film stripes along their should at least locally vary linearly with
in-plane direction temperature. The sample needs to be
stable in the vacuum
3ω method based on In-plane thermal conductivity of films Substrate effect eliminated; The thermal conductance of the bare
the suspended ultrathin Prefabricated metal stripes and insulation layers; membrane needs to be small when
dielectric membranes Compatible with various film deposition techniques; compared to that of the sample
Can obtain κ directly; Possible to be measured with S and σ
SThM-3ω method Thermal conductivity of bulks, thin films, Can obtain κ; Sophisticated models and extensive
or nanostructures Minimum sample preparations; calibrations needed
High spatial resolution;
Applicable in ambient condition
Ångström’s method Thermal diffusivity of bulk materials Easy implementation; Needs Cp and ρ
Simple sample preparations
Temperature wave Cross-plane thermal diffusivity of bulk Easy implementation; Needs Cp and ρ;
analysis (TWA) method materials Simple sample preparations Samples should be thermally thick;
Needs good thermal contact between
sample and heater/sensor
Frequency-domain ther- Thermal conductivity in the Rapid and noncontact; Sophisticated data reduction
moreflectance cross-plane direction, possible in the Applicable to both sub-micrometer thick thin films procedure; May need CV to extract
(FDTR) method in-plane direction and bulk materials; Easier instrumentation than TDTR κ⊥; Large uncertainty for κ∥ when
κ∥ < 5 W m−1 K−1; Smooth sample
surface needed; Needs surface coating
Light irradiated In-plane and cross-plane thermal Easy implementation; Needs Cp and ρ;
ac-calorimetric method ­diffusivity of thin plate samples Possible to measure the in-plane and cross-plane Contact method
thermal diffusivity from the same sample
Modulated photothermal Cross-plane thermal diffusivity and Noncontact method; Complicated normalization
radiometry (PTR) thermal conductivity of bulk materials; Can obtain κ procedures are necessary;
Possible to obtain in-plane thermal Needs surface coating
diffusivity or thermal conductivity of
thin plate-like or filament-like sample.
Photoacoustic (PA) Cross-plane thermal diffusivity and Noncontact method; Needs surface coating;
method thermal conductivity of bulk materials Can obtain κ; Inexpensive detection system; Extensive calibration and
and thin films Toleration for surface roughness normalization procedures
Photo-thermoelectric Cross-plane thermal diffusivity and Easy implementation The sample may need a substantial
method thermal conductivity of bulk materials Can obatain κ; Applicable to thermoelectric liquid Seebeck coefficient

can be detected by the lock-in amplifier. The 3ω voltage signal tance that heat diffuses within the sample during one cycle of
V3ω is given by[131] the ac power heating period[127]) is much larger than the half-
width of the metal stripe b, the metal strip can be treated as a
1
V3ω = I 0 Re,H ∆T2ω β H (23) line source. The complex temperature rise of the metal stripe
2 can be approximated by[127,129,131]
 1 dR 
where βH  βH = e,H
is the temperature coefficient of Pl  1 1  αs  π
 Re,H dT  ∆Ts =  − ln(2ω ) + ln  2  + η − i  (24)
resistance (TCR) of the metal stripe. The thermophysical prop- πκ s 2 2 b 4
erties such as the thermal conductivity of the sample can be
retrieved by fitting proper theoretical transport models to the where κs is the thermal conductivity of the solid and b is the
ΔT2ω data.[127,132] half-width of the metal stripe. Pl is the amplitude of the ac
For a semi-infinite isotropic solid with a thermal diffusivity heating power per unit length dissipated by the metal stripe.
αs, if its thermal penetration depth LP = α s /2ω (i.e., the dis- η is a constant. Equation (24) demonstrates that the thermal

Adv. Electron. Mater. 2019, 1900167 1900167 (13 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

The 3ω method has been extended by Ju et al. to measure


the in-plane thermal conductivity κf,∥ of films by utilizing a very
narrow metal stripe (around several micrometer wide) which is
sensitive to the heat transfer in both in-plane and cross-plane
directions.[133] It can be shown that ΔTf under such a circum-
stance can be written as[74,132]

Pl  1 
1/2
∞ sin 2 λ   df   κ f , 
1/2

∆Tf =
π  κ f ,⊥κ f ,  ∫ λ3
tanh  λ 
 b  
 κ f ,⊥ 
 dλ (27)
 
0

where κf,⊥ is measured separately by using a much wider metal


stripe. In principle, κf,∥ can be readily obtained by fitting the
2D model to the ΔTf data. However, it has been reported by See
et al. in their studies on polymer-nanocrystal hybrid thermoelec-
trics that errors associated with the measurements of the TCR
and the width of the heater line as well as the film thickness
Figure 11. Schematic of the experimental configuration for the traditional propagate highly nonlinearly to cause an uncertainty around
3ω method. 70% for the fitted κf,∥. Therefore, conclusive results for κf,∥ were
not reached in their studies.[134] In fact, for the thermal conduc-
­conductivity of the isotropic substrate κs can be obtained from tivity measurement on the anisotropic ­materials (including the
the slope of the in-phase component of ΔTs as a function of log- anisotropic OHT materials), it is desirable to measure κ⊥ and
arithm frequency. κ∥ independently as in the 2D model the propagation of the
To measure the cross-plane thermal conductivity of an insu- uncertainty in κ⊥ (κ∥) can lead to growth of the uncertainty in
lating layer on the substrate as shown in Figure 11, the width of κ∥(κ⊥) as well.[135] Hence, further comparison of the 3ω method
the stripe is made to be much larger than the thickness of the with alternative techniques may be necessary regarding the
films so that one may assume that the heat flow through the measurement of the in-plane thermal conductivity of the aniso-
film is 1D. Since the film thickness is much smaller than tropic materials.[61]
the thermal penetration depth of the substrate, the measured Up to date the method has been extensively used to
thermal response by the stripe (ΔTsub+f) is just the temperature measure the thermal conductivity of a variety of OHT mate-
rise of the substrate (ΔTsub) added by a frequency independent rials, including the conductive polymer films,[65,66,107,136,137]
temperature rise across the film (ΔTf)[129] the small molecule organic semiconductors,[138–140] the
∆Tsub+f = ∆Tsub + ∆Tf (25) organic/inorganic composites,[134,141,142] etc. The thickness
of the OHT samples prepared for the cross-plane thermal
One may directly calculate ΔTsub from Equation (24), provided conductivity measurements range typically from 100–200
that the thermophysical properties of the substrate have been nm[107,138] to several micrometers.[136,138,142] Among these
determined separately. When ΔTf is known, the cross-plane studies, Bubnova et al. reported a room temperature κ⊥ value
thermal conductivity of the film κf,⊥ is calculated by[129] of 0.33 W m−1 K−1 for the PEDOT:Tos film,[137] which was con-
Pl df sistent with the results measured on similar samples via the
κ f ,⊥ = (26) conventional 3ω[66,107] or TDTR[69,105] techniques. Rausch et al.
2b∆Tf
reported a room temperature κ⊥ value of 0.19 W m−1 K−1 for
where df is the thickness of the thin film. It should be empha- the sub-micrometer thick P3HT film,[143] which is also in good
sized that the thermal contact resistances between the film and agreement with the TDTR[98] and other conventional 3ω[65]
the substrate as well as between the metal stripe and the film measurements.
have been assumed to be negligible here. A major advantage of the 3ω method is that it is insensi-
The thermal conductivity measurement of the TE films often tive to the radiation and convective heat losses, since the
involves multiple-layer structures. For example, for the TE film thermal ­penetration depth associated with the ac heating is
deposited on a SiO2/Si wafer and capped with a SiNx insulation usually small and the heat losses during the measurement
layer, there are totally four layers with different thermophysical occur mainly from a small area surrounding the heating
properties. Theoretically, the temperature rises associated with stripe.[127,144] Therefore, the method can be applied in both
the Si wafer and the SiO2 and SiNx layers can be measured sep- vacuum and ambient. Kim et al. have found that the ther-
arately. But this will often bring about “stacking errors” to the moelectric properties of PEDOT doped by different amounts
final result.[131] One reliable way to subtract the contributions of hydrophilic PSS can be tuned by varying humidity con-
of the substrate and the dielectric layers is known as the “differ- ditions.[66] In their studies, the traditional 3ω method was
ential method,” in which metal stripes with same dimensions employed to perform the thermal conductivity measurements
are patterned onto the specimen and a “blank sample” without in a humidity-controlled environment.
the film of interest.[132] The temperature rises are measured Although the method is relatively easy to implement, the
as a function of the heating frequency with same input power necessity of microfabricating the metal stripe makes the meas-
for both samples. ΔTf is then taken from the temperature-rise urement somehow time-consuming. In some cases, it is also
­difference between the two samples. challenging to fabricate the thin insulation layer with good

Adv. Electron. Mater. 2019, 1900167 1900167 (14 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

quality for isolating the electrically conducting materials to be simultaneously obtain the thermal conductivity and the thermal
measured from the metal stripe.[131,145] To apply the traditional diffusivity (or volumetric capacity) of the sample by performing
3ω method on the OHT materials, more precautions should proper data fittings.
be taken for the sample preparations due to the fact that some On the other hand, when the measurement is carried out at
OHT materials are vulnerable to the relatively high processing the low frequency limit ωγt → 0, V3ω,RMS is almost independent
temperature (e.g., 573 K in a typical plasma enhanced chemical on frequency
vapor deposition (PECVD) process for depositing the insula-
tion layer[13]), and the harsh chemicals used in the lithography
3
4I RMS 2
LRe,s βs
V3ω ,RMS = (29)
procedure. In order to address the issues mentioned above, π κ s As
4

Kaul et al. have developed a sample preparation methodology


which included a low temperature PECVD procedure for fab- In this scenario, the thermal conductivity can be simply found
ricating the insulation layer and utilization of shadow masks by solving Equation (29).
for creating the metal stripe.[13] The recipe has been success- The self-heating 3ω method provides a simple way to
fully applied to the measurements of 110 nm and 5 µm thick measure the thermal conductivity of the electrically conductive
­polyaniline films. materials, including carbon nanotube,[148] inorganic semicon-
ductor nanowires or nanoribbons,[149–151] metal nanowires,[152]
and semiconductor nanocrystal coated glass fibers.[153] In
4.2. Self-Heating 3ω Method recent years, the method has been applied to characterize
highly ­conductive organic/inorganic nanocomposites[154] and
In the self-heating 3ω method,[146,147] the sample itself acts as OHT materials such as poly(nickel-ethylenetetrathiolate)[155]
a heater and temperature sensor, therefore it is required to be and polythiophene[156] films, which were usually several
electrically conductive with a substantial TCR βs. As shown in micrometers to several tens of micrometers thick. In ref. [156],
Figure 12, the measurement is performed with a four-probe Zhang et al. reported a room temperature thermal conductivity
configuration. Two outer electrical contacts are used to feed of 0.22 W m−1 K−1 for a polythiophene sample prepared by
the electrical current, while the electrical voltages are extracted ­electrochemical polymerization with a low electrical conduc-
from two inner electrical contacts. The central portion of the tivity (σ = 44 S cm−1). The result is very close to the bulk value
sample between the two voltage contacts is suspended to avoid of 0.19 W m−1 K−1 measured by the photoacoustic method for
the parasitic heat flow entering into the substrate. The test another electrodeposited polythiophene sample.[60] Huewe
platform is placed in a vacuum chamber to avoid convective et al. have integrated the self-heating 3ω technique into a mul-
heat losses. When a sinusoidal electric current at frequency ω tipurpose sample holder capable of characterizing all three
is passed through the sample, the resistance fluctuation at fre- thermoelectric quantities (S, σ, and κ) on the same sample of
quency 2ω is generated in response to the temperature fluctua- crystalline organic semiconductors.[157] Their studies yielded a
tion at the same frequency. Consequently a small 3ω voltage thermal conductivity of 1.73 W m−1 K−1 at room temperature
signal will be produced which is encoded with information for a DCNQI2Cu (DCNQI = dimethyl-dicyanoquinonediimine)
of the thermophysical properties of the sample. Lu et al. have organic crystal along its axial direction, which agrees in mag-
shown that the amplitude of 3ω voltage signal can be written nitude with literature data for similar quasi-1D organic single
as[146] crystals.[158]
Since the sample itself is used as the thermometer, its resist-
3
4I RMS 2
LRe,s βs ance should vary linearly with temperature, at least within
V3ω ,RMS = (28)
π κ s As 1 + (2ωγ t )2
4
small temperature range around the measurement tempera-
ture. Its electrical and thermal properties need to be stable
where V3ω,RMS and IRMS are the root-mean-square (RMS) ampli- under vacuum condition, especially when the sample is made
tude of the 3ω voltage and the exciting current, respectively. L from the OHT materials. For the OHT materials with humidity-
is the length of the sample between the two inner contacts. κs, dependent electrical and thermal properties,[66] the calibration
Re,s, and As are the thermal conductivity, the resistance, and the of the TCR should be carried out in vacuum environment along
cross-sectional area of the sample, respectively. γt = L2/π2αs is with the 3ω measurement in order to maintain consistency of
a characteristic thermal time constant associated with the heat the data.
transfer along the axial direction, where αs is the thermal dif-
fusivity of the sample. Based on Equation (28), it is possible to
4.3. 3ω Method Based on the Suspended Ultrathin Dielectric
Membranes

The 3ω method can also be coupled with the suspended


ultrathin dielectric membranes to measure the in-plane thermal
conductivity of the films.[44,47,48,65,159,160] One way to achieve this
is to use the λ-chip in the Völklein method (Figure 5a) as the
measurement platform.[44,47,48,159,160] By injecting an ac cur-
Figure 12. Schematic of the experimental setup for the self-heating 3ω rent I(ω) = I0cos(ωt) into the central metal stripe, the gener-
method. ated thermal wave will flow from the heat source toward the

Adv. Electron. Mater. 2019, 1900167 1900167 (15 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

heat sink (i.e., silicon chip) through the suspended membrane The in-plane thermal conductivity is measured for both
(bare dielectric membrane or composite layer composed of the the bare dielectric membrane and the composite structure
dielectric membrane and deposited film). By measuring the 3ω deposited with the film of interest. The thermal conductivity
voltage V3ω across the metal stripe, the 2ω temperature oscil- of the deposited film can be subsequently deduced from
lation ΔT2ω at the location where the stripe is located can be Equation (12).
given by[44] Linseis et al. have applied this method to measure the
temperature-dependent in-plane thermal conductivity of
2V3ω the PEDOT:PSS films with a thickness ranging from sub-
∆T2ω = (30)
I 0 Re,H β H micrometers (100–383 nm)[47] to 15 µm.[48] In ref. [47], the
error bar range associated with the measurement increases
where βH and Re,H are the TCR and electrical resistance of the with decreasing the film thickness and the results for the
metal stripe, respectively. 100 nm thick film demonstrate a high degree of uncertainty.
In order to establish the relationship between V3ω and the The thermal conductivities of the 276 and 383 nm thick
thermophysical properties of the membrane, a 1D transient samples (σ∥ < 20 S cm−1) vary approximately from 0.39 to
heat transfer equation has been solved by imposing proper 0.54 W m−1 K−1 at room temperature, which are comparable
boundary conditions.[159] The amplitude of V3ω in the low fre- with those of the low-σ∥ samples measured by the MFSD[61]
quency limit can be written as[48] and TDTR[69] techniques. On the other hand, the in-plane
thermal conductivity of the 15 µm thick film (σ∥ ≈ 200 S cm−1)
β HRe,H
2
I 03 prepared by drop casting was determined to be ≈0.35 W m−1
V3ω = (31)
4K P K−1 at room temperature,[48] which is smaller than that of the
sub-micrometer thick samples with low σ∥.[47] However, the
2κ  dLM value is very close to the κ⊥ value of the PEDOT:PSS films
KP = (32)
w which has been found to be independent on σ∥.[69] The low κ∥
value of the thick film probably reflects the isotropic nature of
where w and LM are the width and length of the mem- the sample.
brane as defined in Figure 5a. κ∥ and d are the in-plane Ushirokita and Tada have taken another approach to com-
thermal conductivity and the thickness of the membrane, bine the 3ω technique with the suspended ultrathin mem-
respectively. brane.[65] Their test platform is based on a double stripe
It has been found that the values of κ∥ calculated from ­configuration[161–163] as shown in Figure 13. Two metal stripes
­Equations (31) and (32) are subject to errors caused by the (denoted as S1 and S2, respectively) are fabricated on top of a
radiation heat loss from the membrane surfaces.[48] Therefore, SiNx membrane window. A 50 nm thick SiO2 layer is depos-
a correction procedure based on two membranes with different ited on top of the stripes as an insulating layer for subsequent
dimensions has been proposed as follows, from which the deposition of the conducting sample. An ac current in the form
thermal conductivity κ∥ and emissivity ε of the membrane are of I(ω) = I0sin(ωt) is passed through the strip S1 to induce the
obtained together.[48] 2ω temperature oscillation. The heating frequency is selected
It can be shown that the thermal conductance associated to be high enough to make the thermal penetration length
with the two membranes due to heat conduction and heat radi- Lp inside the membrane much smaller than the width of the
ation is written as[44,48] membrane window. Under the 1D heat transfer approximation,
the amplitude of the temperature oscillation at S1 (|ΔTS1|) and
PAC,i  wi 
GM,i = = 2κ  dLM,i µ coth  µ  (33)
∆T2ω ,i  2

8εγ T03
µ2 = (34)
κ d
where PAC,i and |ΔT2ω,i| are the amplitude of the ac heating
power and amplitude of the temperature rise associated with
the ith (i = 1,2) membrane, respectively. γ is the Stefan–Boltz-
mann constant and T0 is the temperature of the heat sink. The
ratio of the thermal conductance of the two membranes is con-
sequently given by[48]

LM,1 coth  µ 1 
w
GM,1  2
= (35)
LM,2 coth  µ 2 
GM,2 w
 2

By numerically solving Equation (35), the parameter μ can be


obtained readily. It is then straightforward to calculate κ∥ and ε Figure 13. The platform for measuring the thermal conductivity of a thin
from Equations (33) and (34). film using double metal stripes.

Adv. Electron. Mater. 2019, 1900167 1900167 (16 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

the phase shift of the temperature oscillation at S2 (ΔϕS2) are ink-jet printing).[48] On the other hand, to ensure a sufficient
given by[65,161] measurement sensitivity, the thermal conductance of the test
film should be as large as possible when compared to that of
∆TS1 α 1 the bare membrane.[65] In contrary to the conventional 3ω tech-
= (36)
Pl 2 2dκ  ω nique, this method should be conducted under the vacuum
environment, which precludes its use in exploring the thermal
and conductivity of the OHT materials in ambient or humidity-con-
trolled conditions.[66]
ω
∆ϕS2 = − L + ϕ0 (37)
α
4.4. SThM-3ω Method
respectively. Here, Pl is the amplitude of the ac heating power
per length. κ∥ and α∥ are the in-plane thermal conductivity and SThM brings a microprobe which acts as thermometer and
thermal diffusivity of the suspended membrane, respectively. heater to be in contact with or in close proximity to the sample
d is the thickness of the membrane. L is the interval between surface.[165–167] By delivering heat locally to the sample and
the two stripes and ϕ0 is a constant. |ΔTS1| is derived from the measuring corresponding thermal responses, it is possible to
­measurement of the 3ω voltage. To obtain ΔϕS2, a dc “probe” probe the thermal conductivity of the sample with high lateral
current is passed through S2 to detect its resistance vari- resolution. More specifically, the probe could be a thermo-
ation. Since the temperature rise of S2 has a 2ω component, couple junction[168,169] or a resistive metallic probe.[170–174] In
the resistance of S2 gains a small ac component of the same the latter case, the temperature variation is detected by meas-
frequency. The product of the ac resistance with the “probe” uring the resistance change of the probe. If an ac current with
current gives rise to a small voltage component oscillated at angular frequency ω is used to heat the resistive metallic probe,
2ω.[162,164] The phase shift of this small voltage signal is ΔϕS2 the resulted 2ω temperature oscillation of the probe should be
which can be measured by the lock-in amplifier. In the experi- proportional to its voltage rise in the third harmonic (i.e.,V3ω)
ment, both ΔϕS2 and |ΔTS1| are recorded as a function of ω. α∥ that can be detected by a lock-in amplifier. An experimental
can be obtained by linearly fitting ΔϕS2 as a function of ω . setup adopted by Lefèvre and Volz for performing the SThM-3ω
With known α∥, κ∥ is derived from the slope of the |ΔTS1| versus measurement is schematically shown in Figure 14.[172]
ω−1/2 curve. Based on this configuration and Equation (12), the In order to link the observable quantity V3ω with the thermal
thermal conductivity of poly(methyl methacrylate) (PMMA) and conductivity, a number of analytical and numerical models
several thin conjugated polymer films (200 nm–1 µm thick) have been proposed, which take into accounts of various factors
have been characterized at room temperature by Ushirokita and such as the geometrical/dimensional parameters and mate-
Tada.[65] The room temperature value of the PMMA film was rials properties of the probe, the effective thermal resistance
found to be 0.17 W m−1 K−1, which is in good agreement with between the sample and the probe, the effective coefficient of
the literature result.[88] They also reported a correlation between heat losses by the whole probe surface to its environment, and
κ∥ and the in-plane electrical conductivity σ∥ for the PEDOT:Tos so on.[171,172,175–178]
films, which confirms the previous observations made on the The technique allows relatively fast characterization of a
PEDOT:PSS and PEDOT:Tos samples by applying the MFSD[61] wide variety of sample types (including bulks, films, or nano-
and TDTR[69] techniques. All these studies actually yield close structures)[166,172,176,178–181] and requires only minimal sample
κ∥ values (around 0.5–0.7 W m−1 K−1) for the low-σ∥ samples preparations. It provides a high lateral spatial resolution[166]
(σ∥ < 50 S cm−1).[61,65,69] The κ∥ value for the sample with that is not attainable by other techniques. For the bulk or ther-
σ∥ = 240 S cm−1 was found to be 0.86 W m−1 K−1 in ref. [65], mally thick films, the thermal conductivity can be derived by
which is coincident with the result (≈0.82 W m−1 K−1 at treating the sample as a semi-infinite medium. If the film on
σ∥ ≈ 250 S cm−1) obtained via the TDTR technique[69] and is a ­substrate is thermally thin and heat spreading in the in-plane
about 28% smaller than the value interpolated from the MFSD directions is non-negligible, a 2D heat transfer model for a
data (≈1.2 W m−1 K−1).[61] multilayered system should be employed to obtain the thermal
To summarize, the 3ω method based on the suspended
ultrathin membrane can be used to measure the in-plane
thermal conductivity of the films with various thicknesses of
sub-micrometers to more than ten micrometers by eliminating
the substrate effect. The chip-based measurement platform
has been designed to measure the Seebeck coefficient and
the electrical conductivity of the sample at the same time.[47]
Unlike the conventional 3ω method, this method utilizes the
prefabricated insulation layers and metal stripes, which is an
advantage for characterizing the OHT materials that may not
be able to sustain hostile processing conditions. The test-chip
is designed to allow facile sample deposition by PVD methods
(e.g., thermal evaporation and magnetron sputtering) and Figure 14. Schematic of the experimental setup for the SThM-3ω
solution-based techniques (e.g., spin coating, drop casting, or ­measurement adopted by Lefèvre and Volz.[172]

Adv. Electron. Mater. 2019, 1900167 1900167 (17 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

c­ onductivity.[176] In recent years, the method has been used to characterizing the thermal transport properties of the bulk
measure the thermal conductivity of several inorganic nano- thermoelectric materials, the Ångström’s method still remains
structures such as the nanowires of Si,[178,179] SiGe,[179] and useful due to its simplicity and versatility. Recently, Shin
Bi2Te3,[180] and some OHT materials including P3HT nanow- et al. have used this method to characterize the Bi0.5Sb1.5Te3
ires (100 µm long and 120–350 nm in diameter),[181,182] doped and Bi2Te2.7Se0.3 based samples prepared by a combination of
and undoped poly[N-9′-heptadecanyl-2,7-carbazole-alt-5,5-(4′,7′- screen-printing and hot pressing procedures.[191] Tan et al. have
di-2-thienyl-2′,1′,3′-benzothiadizole)] (PCDTBT) films (several used this method to measure the thermal diffusivity of a kind
micrometers thick),[176,183] and so on. In order to evaluate the of carbon nanoparticle hybrid aerogel with extremely low den-
reliability of the SThM-3ω technique, Wilson et al. carried out sity and high porosity.[124] The derived thermal conductivity for
characterizations on calibrated samples, including PANI com- the aerogel sample was about 0.02–0.03 W m−1 K−1, which is
posites mixed with 5% and 7% graphene nanoplatelets,[176] comparable with that of a 3D CNT network measured via the
which have been investigated previously by the comparative laser flash and hot disk methods at low air pressure (20 Pa).[126]
method.[184] The measured thermal conductivities were found The main limitation of the Ångström’s method is that it does
to be 0.47 and 0.68 W m−1 K−1, respectively, which are in good not directly yield the value of κ, although certain attempt has
agreement with the calibrated data. Since the SThM measure- been made to modify the method in order to achieve a simulta-
ments are usually performed in air condition, the method is neous measurement of α and κ.[186]
in principle applicable to the investigations of the humidity
dependent thermal conductivity[66] of the OHT ­materials. On
the other hand, in order to extract the value of the thermal con- 4.6. Temperature Wave Analysis Method
ductivity, the sophisticated models and extensive calibrations
are generally needed. The accuracy of the results relies on how The temperature wave analysis (TWA) method is a contact
those effective parameters associated with the theoretical model method for measuring the cross-plane thermal diffusivity of the
have been obtained.[171,172,175–178,185] sample.
In this method, the sample of thickness d is sandwiched
between one heater and one resistive temperature sensor[192–194]
4.5. Ångström’s Method (Figure 16). The heater creates a thermal wave at the front sur-
face of the sample by the ac Joule heating at frequency f mod
The Ångström’s method is a widely accepted method for deter- ( f mod = ω mod /2π ) , which diffuses toward the rear surface
mining the thermal diffusivity of the solid materials.[7,186,187] In along the thickness direction. The temperature oscillation at
this method, the sample is heated periodically at one end, while the rear surface is measured by detecting the resistance varia-
the temperatures are measured at two locations along its length tion of the sensor with the lock-in technique. If the sample is
(Figure 15). When the propagation of the thermal wave reaches thermally thick ( i.e., d π f mod /α  1), the phase lag of the tem-
a steady state, temperature variations are recorded to calculate perature variation between the front surface and the rear sur-
the thermal diffusivity by[188] face, Δϕ, is approximated by[193,194]

π f mod L2 ω mod π
α= (38) ∆ϕ = − d− (39)
∆ϕ ln ( ∆Th / ∆Tc ) 2α 4

where f mod is the heating frequency. L is the distance between The thermal diffusivity α is determined from the slope of Δϕ
xh and xc where two temperature sensors are located. |ΔTh| and versus ω mod plot.
|ΔTc| are the measured amplitudes of the temperature wave at The TWA method only needs simple instrumentation and
xh and xc, respectively, while Δϕ is the phase lag between the minimum sample handlings. The major limitation of the
two locations.
The Ångström’s method is a traditional method for obtaining
the thermal diffusivity of the rod-like or stripe-like sample.
The implementation of the method is relatively easy and the
sample preparations are simple. It has been used to measure
the thermal diffusivity of the thermoelectric materials since
1960s.[188–190] Though many choices are available today for

Figure 16. The experimental setup for the temperature wave analysis
Figure 15. The experimental configuration for Ångström’s method. method.

Adv. Electron. Mater. 2019, 1900167 1900167 (18 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

change with the probe laser beam.[199,200]


However, in contrast to the TDTR method
in which the thermoreflectance change is
measured as a function of the delay time
between the probe beam and the pump
beam, the FDTR method records the output
signals as a function of modulated frequency.
Both the ultrafast pulsed laser and CW laser
can be used in this technique. Accordingly,
two major experimental setups have been
proposed by Schmidt et al. for carrying out
the FDTR measurements.[199] The first kind
of experimental setup is converted from
Figure 17. Schematic of the FDTR system based on CW lasers proposed by Schmidt et al.[199] the TDTR system similar to the one shown
in Figure 9b by fixing the position of the
method is that one has to know the specific heat and density mechanical delay stage and varying the modulation frequency.
of the sample in order to calculate the thermal conductivity. The second type of experimental setup is implemented by using
To ensure measurement reliability, the sample should be ther- the CW lasers as the pump and probe beams (Figure 17). In
mally thick and good thermal contacts should be achieved at this case, the pump laser is externally modulated and the probe
the sample/heater and the sample/sensor interfaces. The tech- beam is directed at the sample surface without passing through
nique has been previously applied to measure the temperature the mechanical delay stage. Therefore, the setup can be built
dependent cross-plane thermal diffusivity of nonconductive in a less-expensive and less complicated way when compared
polymer films, including aromatic polyolefins[193] and poly- with the FDTR setup based on the ultrafast laser system and
imide.[194] Morikawa and Hashimoto have measured the tem- the mechanical delay stage. To obtain the thermophysical prop-
perature-dependent thermal diffusivity for 25 and 125 µm thick erties, the amplitude and phase shift of the thermoreflectance
polyimide (Kapton) films,[194] which demonstrated a good coin- signals recorded by the lock-in amplifier should be analyzed
cidence with the laser flash results.[195] with proper heat transfer models.[74,85,199] In practice, the volu-
Later, its application scope has been extended to OHT mate- metric heat capacity CV of the sample is often used as one of
rials. With this technique, Hiroshige et al. investigated the known input parameters to the models for deriving the thermal
thermal diffusivity of pristine and iodine-doped copolymers of conductivity value.[135,201,202]
phenylenevinylene with dialkoxyphenylenevinylene.[196] Zhang The measurement sensitivity of the FDTR method is com-
et al. have measured the thermal diffusivity of PEDOT:S-PHE pared with that of the TDTR method[199] and both methods
(sulfated poly(β-hydroxyethers)) films prepared by electro- should be applicable to the small samples with dimensions
chemical polymerization.[197] The room temperature thermal down to the order of 100 × 100 × 100 µm3.[201] Since the
diffusivity of these thermoelectric polymers was found to be thermal penetration depth of the FDTR method ranges typi-
around 1–3 × 10−7 m2 s−1, which is of the same order as that cally from 200 nm to several micrometers at a modulation
of the insulating Kapton films (≈1.2 × 10−7 m2 s−1).[194] By pre- frequency of 0.1–20 MHz, it is also regarded as a powerful
paring 0.74–0.84 mm thick SWCNT–polystyrene composites, technique for probing the cross-plane thermal conductivity
Suemori and Kamata were able to measure the thermal diffu- of the nanoscale thin films.[86] Ong et al. have used FDTR to
sivity along both cross-plane and in-plane directions by proper measure the thermal conductivity a series of novel mate-
orientating the sample slices.[198] The derived thermal conduc- rials such as ligand-stabilized 3D inorganic nanocrystal
tivities along the in-plane direction for the composites with 25 arrays (NCAs) layer (30–150 nm thick)[202] and supera-
and 50 wt% SWCNTs are around 4 W m−1 K−1, which are about tomic crystals (SACs).[201] These materials are expected to
5 times and 10 times larger than those along the cross-plane serve as electronic, thermoelectric, and photonic materials
direction due to the preferable orientation of SWCNTs in the in the future. It was found that the thermal conductivities
in-plane direction. The in-plane thermal conductivity is actually of the NCAs and SACs are typically within a range of
comparable with the axial thermal conductivity of the aligned 0.1–0.3 W m−1 K−1 at room temperature. The FDTR technique
CNT-thermoset epoxy composite with 16.7 vol% CNTs meas- was able to reveal that the thermal conductivity of the NCAs
ured by Marconnet et al. using the comparative method increases with increasing diameter of the nanocrystal core,
(4.87 W m−1 K−1).[37] but is less sensitive to the thermal conductivity of the core
materials.[202] Malen et al. applied a fiber-aligned FDTR tech-
nique to measure the room-temperature thermal conductivity
4.7. Frequency-Domain Thermoreflectance Method of a 1.6 µm thick P3HT film.[203] The value was found to be
0.17 W m−1 K−1 which is in agreement with that obtained by the
The FDTR technique is a rapid and noncontact method for conventional 3ω[65] or TDTR[98] methods.
measuring the thermal conductivity of both sub-micrometer The FDTR method is usually involved with the complicated
thick films and bulk materials. Like the TDTR method, the data reduction procedures. It also requires smooth sample
FDTR method also uses the modulated pump laser to periodi- surface[14] and application of the metallic coating. Although
cally heat the sample surface and detects the thermoreflectance the method has been modified to characterize the in-plane

Adv. Electron. Mater. 2019, 1900167 1900167 (19 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

thermal conductivity of the anisotropic materials by using the


beam-offset configuration, large uncertainty for measuring
the in-plane thermal conductivity below 5 W m−1 K−1 has been
observed.[135] This might bring difficulties to the reliable meas-
urement of the in-plane thermal conductivity of the OHT mate-
rials which is usually less than 5 W m−1 K−1.[47,61,63,64,67]

4.8. Light Irradiated ac-Calorimetric Method

In the light irradiated ac-calorimetric method, the sample is


periodically irradiated by a modulated light source and the
thermal response is measured by a thermocouple.[204,205] There-
fore, this method is essentially a contact method.
As will be shown later, the technique can be implemented in
Figure 18. Schematic of the experimental configuration for the light
several configurations. The primitive configuration proposed by ­irradiated ac-calorimetric method proposed by Hatta et al.[204]
Hatta et al. in 1980s was aimed to measure the in-plane thermal
diffusivity of the thin plate samples.[204] In this configuration, a junction in some studies is formed between a sharp metal tip
modulated light source is used to irradiate the sample which is and the metal layer coated on the backside of the sample, the
partially blocked by a movable screen. The temperature oscilla- measurement technique in this case was thus named as photo-
tion on the rear side of the sample is monitored by a thermo- thermoelectric method in order to emphasize the thermoelec-
couple (Figure 18). If the thickness d of the sample is assumed tric nature of the temperature sensor.[209,210,212] It should not
to be much smaller than the thermal penetration depth, the be confused with the photo-thermoelectric method that will be
temperature gradient can be neglected in the film thickness mentioned in Section 4.11, in which a transient thermoelectric
direction. The heat flow is then viewed as a 1D diffusion pro- voltage generated from the sample itself is used to measure the
cess along the x-direction under the shadowed region. The ac temperature rise.
temperature rise measured by the thermocouple is related to In summary, the experimental apparatus for the light irra-
the distance x between the movable screen edge and tempera- diated ac-calorimetric method is relatively easy to implement.
ture sensing location through[204] By employing the method, it is even possible to measure the
in-plane and cross-plane thermal diffusivity from the same
QA
∆T ( x ) = e − kx − i (kx + π /2) (40) sample.[208] However, in order to calculate the thermal conduc-
2ω mod C V d tivity, the density and the specific heat should be determined
separately. Additional coating on the sample surface may be
where ω mod(ω = 2π f mod ) is the angular modulation frequency, necessary to enhance the light absorption.[208] The contact
k is the inverse thermal penetration length in the form of resistance between the sample and the temperature sensor
k = π f mod /α , QA the absorbed power density (W m−2). The (e.g., thermocouple) will also bring errors to the results.[210]
thermal diffusivity can be retrieved from the amplitude |ΔT(x)| The method has been employed to characterize the in-plane
or phase ϕ of the temperature oscillation as a function of the thermal diffusivity of several conducting polymer or organic
screen offset x[206] composite materials, including polypyrrole films (less than
120 µm thick),[213] nanofibrillated cellulose (NFC)/PEDOT:PSS
π f mod based mixed ion-electron conductor (30 µm thick),[214] NFC/
αθ = (41)
(d ln ∆T ( x ) /dx )2 polystyrene sulfonate sodium based thermoelectric composite
(30 µm thick),[215] etc. In addition, Zhang et al. measured the
π f mod thermal diffusivity of a few millimeter-sized crystalline organic
αϕ = (42)
(dϕ /dx )2 semiconductors which have a transverse thickness in the range
of 300–610 µm.[158,211] The derived thermal conductivity along
where αθ and αϕ are the thermal diffusivity values extracted growth axis for a needle- shaped 6,13-bis(triisopropylsilyleth
from the amplitude and the phase signal, respectively. ynyl) pentacene crystal was found to be 1.6 W m−1 K−1 at room
The method was later modified by replacing the uniform temperature,[158] which is comparable with that of the quasi-1D
illumination by scanning line-shape and spot-like laser beam organic crystals consisting of segregated molecular stacks with
on the sample surface.[205,207–210] Yu et al. have compared these excellent p-orbital overlaps.[157]
three configurations and suggested that the uniform illumina-
tion method is more susceptible to error than the other two
configurations.[206] On the other hand, the cross-plane thermal 4.9. Modulated Photothermal Radiometry
diffusivity can be obtained if the spot-like light source which
excites the front surface of the sample is aligned with the In the modulated photothermal radiometry (PTR), the
thermocouple on the opposite side of the sample,[208,209] or if sample is illuminated by an intensity modulated light beam.
the whole front surface of the sample is homogenously irra- A monolithic Infrared detector coupled with a lock-in ampli-
diated.[158,211] It should be mentioned that the thermocouple fier is used to sense the temperature oscillation which can

Adv. Electron. Mater. 2019, 1900167 1900167 (20 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

A linear fit of the phase (or the natural loga-


rithm of the amplitude) of Tn against f mod
results in a straight line with a slope of m
from which the thermal diffusivity is calcu-
lated as[219]
2
d
α =  π (46)
m

On the other hand, when the sample is semi-


transparent, both α and αv can be simulta-
neously obtained by fitting the experimental
data to Equation (44).
Figure 19. Schematic sketch of the experimental setup for modulated photothermal radiometry.
In contrast to the rear-side detection con-
a) Front-side detection configuration. b) Rear-side detection configuration.
figuration, the front-side detection configura-
tion can be used to simultaneously retrieve
be used to deduce the thermal transport properties of the the thermal diffusivity and effusivity of the sample.[218,221,222] Its
sample.[216–219] normalization procedure involves with getting the ratio of the
This technique can be applied in either the front-side detec- frequency dependent PTR signals collected from the sample
tion configuration (Figure 19a) or the rear-side detection config- with the backing substrate and without the backing substrate.
uration (Figure 19b) depending on whether the thermal emis- When the substrate is thermally thick, the normalized signal Sn
sion is recorded from the irradiated surface or from the oppo- is expressed as[218]
site-side surface of the sample. The rear-side detection configu-
ration is often used to characterize the thermal diffusivity of the 1 + (Ysb − 1)exp( −2q k d )
Sn = (47)
solid materials. [219,220] Assuming that a sample with thickness 1 − (Ysb − 1)exp( −2q k d )
d is homogeneously heated by a monochromatic light beam of
intensity Iv0 at a modulation frequency ω mod(ω = 2π f mod ) from Here, Ysb is the reflection coefficient of the thermal waves at
the front surface, the light intensity at a certain position inside the interface between the sample and the substrate. It can be
the sample is expressed as I v ( x ) = I v 0 (1 − R v )e −α v x, where x is the written as[218]
distance from the front surface to this location. Rv and αv are
the optical reflection coefficient and the absorption coefficient Ysb = (es /e b ) − 1 (48)
of the sample at the wavelength of the incident light, respec- (es /e b ) + 1
tively. By neglecting heat losses, the ac component of the tem-
perature rise inside the sample is given by[219] where es and eb are the thermal effusivities of the sample and
the substrate, respectively. It can be further shown that the
I v 0 (1 − R v )α v2  (e −α d − e − q d )e q x + (e −α d − e q d )e − q x + (q k /α v )(e q d − e − q d )e −α x 
v k k v k k k k v
thermal diffusivity of the sample is expressed as[218]
∆T ( x ) =  
2κ q k (q k2 − α v2 )  e qk d − e − qk d 
(43) 4 f mod ,0 d 2
α= (49)
π
where q k = iω mod /α is the thermal wave vector, κ and α are
the thermal conductivity and thermal diffusivity of the sample, where f mod ,0 corresponds to the smallest frequency where the
respectively. phase of Sn becomes zero. Once α has been obtained, Ysb can
In the PTR method, in order to eliminate the instrumental be calculated from the phase of the Sn data. Since eb is a known
influence, normalization procedures are needed. One way of value, one may directly calculate es by Equation (48)[218] and the
achieving this is to normalize the signals detected at the rear thermal conductivity of the sample by Equation (4).
surface ΔT(d) to those recorded at the front surface ΔT(0)[219] If the infrared detector is replaced by a fast infrared video
camera with a high lateral resolution, the technique is termed
∆T (d ) (e −α v d − e − qk d )e qk d + (e −α v d − e qk d )e − qk d + (q k /α v )(e qk d − e − qk d )e −α v d as lock-in infrared thermography which can be employed to
Tn = =
∆T (0) 2e −α v d − e qk d − e − qk d + (q k /α v )(e qk d − e − qk d ) obtain the in-plane thermal diffusivity or thermal conductivity
(44) of thin plate-like or filament-like samples.[223–226]
In summary, the modulated PTR technique is a well-­
where Tn is the normalized temperature. established noncontact method for obtaining the thermal
Equation (44) suggests that Tn does not rely on κ, but diffusivity and thermal conductivity of the bulk materials.
­
depends on d/ α and αvd. If one further assumes that the However, the implementation of the technique involves the
sample is opaque (αvd → ∞) and thermally thick d/ α → ∞ , ( ) complicated normalization procedures.[218,219] To enhance the
Equation (44) is reduced to[219] light absorption, the surface coating is usually applied, which
may restrict the reuse of the sample for other test purposes. In
π f mod π f mod very recent years, the method has been employed to measure
− d −i d (45)
Tn ≈ 2e − qk d = 2e α
e α
the thermal conductivity of bulk PANI,[227] PANI/reduced

Adv. Electron. Mater. 2019, 1900167 1900167 (21 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

graphene oxide composite,[227] and PANI/silver composite,[228] Experimentally, the PA signal can be recorded with the front-
as well as the transverse thermal diffusivity of millimeter-sized side detection configuration (Figure 20a) and the rear-side
organic semiconductors (up to 600 µm thick in the transverse detection configuration (Figure 20b). In order to eliminate the
direction).[229] The thermal conductivity of the PANI sample frequency dependence of the instrumental responses and non-
could be used as a benchmark for verifying the reliability essential signal-related proportionality factors, normalization
of the measurements. With the PTR technique. Islam of the PA signals is typically needed. For example, in order to
et al.[227] reported a room temperature thermal conductivity of measure the thermal diffusivity of the solids, several normaliza-
≈0.44 W m−1 K−1 for a PANI sample with an electrical conduc- tion procedures have been proposed. Pessoa et al. performed
tivity of less than 0.1 S cm−1, which is consistent with that of the normalization by taking the signal’s ratio in the rear-side
the low-σ sample (0.46 W m−1 K−1 at σ = 0.48 S cm−1) meas- and front-side detection configurations for one-single layered
ured by the comparative method by Abad et al.[184] The result sample at a single modulation frequency.[234] Balderas-López
is also comparable with the cross-plane thermal conductivity took the signal ratio from the investigated sample and a refer-
of 0.59 W m−1 K−1 measured by the conventional 3ω tech- ence sample which were both measured in the rear-side detec-
nique for a 5 µm thick PANI film with an electrical conduc- tion configuration.[235] Alternatively, the test sample can be
tivity of 93 S cm−1.[13] measured in the rear-side detection configuration, while the
reference sample in the front-side configuration.[236]
In principle, the thermal conductivity of the bulk sample can
4.10. Photoacoustic (PA) Method be determined by, for instance, putting the sample in thermally
contacted with a backing materials whose thermophysical prop-
In the PA method,[230–232] an intensity modulated radiation erties have been known.[237] On the other hand, by performing
heats a sample located in a closed gas cell (PA cell). A gas layer the measurements at a frequency up to 20 kHz and applying
in contact with the sample surface is heated periodically and the advanced multilayer model[233] for data reduction, the tech-
is driven to expand and contract like a thermal piston. The nique has been extended to measure the thermal conductivity
periodic change of pressure produces an acoustic wave that is of the thin film with sub-micrometer thickness.[238,239]
detected by a microphone, which can be used to derive the ther- The PA technique is a noncontact method featured with
mophysical properties of the sample. a unique and inexpensive detecting system. It can actually
The PA effect was first theoretically explained by Rosencwaig measure the cross-plane thermal conductivity of the films and
and Gersho who revealed that the PA signal (or the variations the bulk samples with rough surfaces.[238] The toleration of the
of the gas pressure) is given by[230] method for the surface conditions may allow one to measure
the thermal conductivity of those OHT materials whose sur-
γ g µgPg,0θ i(ω mod t − π4 ) faces are so rough that the use of the techniques such as the
∆Pg (t ) = e (50)
2LgT0 conventional 3ω method and the TDTR (or FDTR) method
should be precluded. A disadvantage of this technique is that
where Pg,0 is the static pressure and ω mod is the angular modu- it needs extensive calibration and normalization procedures
lation frequency. μg(µg = 2α g /ω mod , where αg is the thermal to eliminate the PA signal distortions caused by the electronic
diffusivity of the gas) and γg are the thermal diffusion depth circuitry.[234–236,238,240] Additionally, the surface coating for the
and the ratio of specific heats of the gas, respectively. T0 is sample is needed to enhance the laser power absorption, which
the average temperature. Lg is the gas column length. θ is a will limit the use the sample for other purposes. To date, the
complex expression for the temperature oscillation at the gas/ method has been utilized to measure the thermal conductivity
sample interface. of several OHT materials including undoped PCDTBT film,[183]
The theory was soon extended by McDonald and Wetsel Bi2Te3 nanowire array/SU-8 epoxy resin nanocomposite,[241]
who suggested that mechanical vibration of the sample sur- poly(3-octylthiophene-2,5 diyl) (P3OT) film and MWCNT/P3OT
face due to thermal expansion exerts work and subsequently composites,[242] polythiophene film and nanofibre arrays,[60]
causes additional temperature change on the gas medium.[231] as well as P3HT nanotube array,[243] etc. The thickness of the
Recently, Hu et al. proposed a generalized expression for the OHT samples usually ranges from several micrometers[183] to
PA effect in a multilayer system.[233] several tens of micrometers.[241–243] Among these materials,

Figure 20. Schematic sketch of the photoacoustic measurement setup. a) Front-side detection configuration. b) Rear-side detection configuration.

Adv. Electron. Mater. 2019, 1900167 1900167 (22 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

the P3OT film was found to have a thermal conductivity coefficient in order to achieve a high measurement accuracy,
of 0.21 W m−1 K−1,[242] which is in agreement with the values although the samples with a Seebeck coefficient as low as
reported for structurally similar polymers such as P3HT.[98,143] 5.8 µV K−1 were still measurable.[244]
The value of the undoped PCDTBT film was measured to be
0.20 W m−1 K−1, which is in reasonable agreement with the result
obtained from the SThM-3ω measurement (0.25 W m−1 K−1).[183] 5. Summary
This paper presents a brief review on the commonly used and
4.11. Photo-Thermoelectric Method newly developed techniques that can be used for characterizing
the thermal conductivity of the OHT materials. Each individual
The photo-thermoelectric method was proposed by Kuria- technique mentioned above has its own merits and l­imitations.
kose et al.[244,245] who used the transient photo-thermoelectric A suitable method is selected primarily based on the factors such
response of the sample as a probe to explore its own thermo- as the characteristics of the sample (e.g., dimensionality, size,
physical properties. The experimental setup is schematically morphology, mechanical properties, optical properties, thermal
illustrated in Figure 21. The top and bottom surfaces of the ther- properties, etc.), the temperature range of the measurement,
moelectric sample are covered with thin metal electrodes for the required measurement accuracy, and so on. When a specific
extracting the thermoelectric voltage. The bottom of the sample method has been selected, it is crucial to understand its theo-
is thermally contact with a substrate with known thermal dif- retical backgrounds and possible error sources. Careful sample
fusivity and thermal effusivity. The top surface of the sample is preparations or instrumentation calibrations are also important
heated by a modulated laser beam. At a fixed modulation fre- for carrying out reliable measurements. To further ensure the
quency, the thermoelectric signal ΔVth obtained with presence accuracy and creditability of the results, some researchers prefer
of the substrate is normalized by the signal ΔVth,0 obtained with to crosscheck their results by different techniques.[126,183,244]
air as the backing material[244] With the continuing development of the novel OHT materials,
some measurement techniques may receive more attention in
∆Vth the future. For example, the characterization of the ­anisotropic
Vn = (51)
∆Vth,0 thermal conductivity has become an important topic in the OHT
community. The combined use of the relevant tools (such as the
where Vn is the normalized signal. By fitting Vn to a theoretical various 3ω techniques, the thermoreflectance ­techniques, the
model,[244] both thermal diffusivity α and thermal effusivity e light irradiated ac-calorimetric method, the modulated photo-
of the sample can be calculated. By applying Equation (4), the thermal radiometry, etc.) might be helpful for tackling the issue.
thermal conductivity κ of the sample is determined. On the other hand, it has been shown in many studies that the
The method provides a simple and rapid way to measure thermal conductivity can be simultaneously measured with the
the thermal diffusivity and thermal conductivity for the bulk Seebeck coefficient and the electrical conductivity on the same
thermoelectric sample. The OHT materials such as 6.6 wt% sample.[22,61,157] This feature is highly desirable, since calculating
oxidized SWNTs/doped PANI composite have been tested the ZT value from the TE properties measured on different sam-
by this method. Its thermal conductivity was measured to be ples may cause errors. This is due to the fact that the samples
0.83 W m−1 K−1, which agrees well with that obtained from the are usually more or less inhomogeneous even when they are
conventional photothermal radiometry.[244] Another interesting chosen from the same batch.[34]
feature of this method is that it can be adapted to measure the
thermal conductivity of thermoelectric liquids, such as tetrado-
decylammonium nitrate salt/1-octanol mixture.[245] It should be Acknowledgements
noted that the sample should possess a relatively large ­Seebeck
The authors would like to acknowledge the financial supports from
the National Natural Science Foundation of China under grant nos.
51573190 and 61176083, and the State Key Development Program for
Basic Research of China under grant no. 2011CB932801.

Conflict of Interest
The authors declare no conflict of interest.

Keywords
hybrid thermoelectric materials, organic thermoelectric materials,
thermal conductivity measurement

Received: February 13, 2019


Figure 21. Schematic illustration of the photothermoelectric experimental Revised: May 15, 2019
setup proposed by Kuriakose et al.[244] Published online:

Adv. Electron. Mater. 2019, 1900167 1900167 (23 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

[1] Q. Zhang, Y. Sun, W. Xu, D. Zhu, Adv. Mater. 2014, 26, 6829. [37] A. M. Marconnet, N. Yamamoto, M. A. Panzer, B. L. Wardle,
[2] Y. Chen, Y. Zhao, Z. Liang, Energy Environ. Sci. 2015, 8, 401. K. E. Goodson, ACS Nano 2011, 5, 4818.
[3] C. Gao, G. Chen, Compos. Sci. Technol. 2016, 124, 52. [38] D. T. Verebelyi, Rev. Sci. Instrum. 1997, 68, 2494.
[4] B. Russ, A. Glaudell, J. J. Urban, M. L. Chabinyc, R. A. Segalman, [39] H. Yoshino, G. C. Papavassiliou, K. Murata, J. Therm. Anal. Cal-
Nat. Rev. Mater. 2016, 1, 16050. orim. 2008, 92, 457.
[5] G. Chen, W. Xu, D. Zhu, J. Mater. Chem. C 2017, 5, 4350. [40] H. Yoshino, G. C. Papavassiliou, K. Murata, Synth. Met. 2009, 159,
[6] A. Zevalkink, D. M. Smiadak, J. L. Blackburn, A. J. Ferguson, 2387.
M. L. Chabinyc, O. Delaire, J. Wang, K. Kovnir, J. Martin, [41] H. Yoshino, H. Aizawa, K. Kuroki, G. C. Anyfantis,
L. T. Schelhas, T. D. Sparks, S. D. Kang, M. T. Dylla, G. J. Snyder, G. C. Papavassiliou, K. Murata, Phys. B 2010, 405, S79.
B. R. Ortiz, E. S. Toberer, Appl. Phys. Rev. 2018, 5, 021303. [42] F. Völklein, Thin Solid Films 1990, 188, 27.
[7] R. Taylor, in CRC Handbook of Thermoelectrics (Ed: D. M. Rowe), [43] F. Völklein, E. Kessler, Measurement 1987, 5, 38.
CRC Press, Boca Raton 1996, Ch. 16. [44] F. Völklein, H. Reith, A. Meier, Phys. Status Solidi A 2013, 210,
[8] Y. Liu, M. Zhang, A. Ji, F. Yang, X. Wang, RSC Adv. 2016, 6, 106.
48933. [45] V. Linseis, F. Völklein, H. Reith, P. Woias, K. Nielsch, J. Mater. Res.
[9] S. Imai, K. Adachi, IEEE Trans. Electr. Electron. Eng. 2015, 10, 101. 2016, 31, 3196.
[10] C. Stacey, A. J. Simpkin, R. N. Jarrett, Int. J. Thermophys. 2016, 37, [46] V. Linseis, F. Völklein, H. Reith, P. Woias, K. Nielsch, J. Electron.
107. Mater. 2018, 47, 3203.
[11] S.-Y. Fu, in Polymers at Cryogenic Temperatures (Eds: S. Kalia, [47] V. Linseis, Z. M. Hassan, H. Reith, J. Garcia, K. Nielsch,
S.-Y. Fu), Springer, Berlin, Heidelberg 2013, Ch. 2. H. Baumgart, E. Redel, P. Woias, Phys. Status Solidi A 2018, 215,
[12] C. Tseng, M. Yamaguchi, T. Ohmori, Cryogenics 1997, 37, 305. 1700930.
[13] P. B. Kaul, K. A. Day, A. R. Abramson, J. Appl. Phys. 2007, 101, [48] V. Linseis, F. Völklein, H. Reith, K. Nielsch, P. Woias, Rev. Sci.
083507. Instrum. 2018, 89, 015110.
[14] P. Q. Jiang, X. Qian, R. G. Yang, J. Appl. Phys. 2018, 124, 161103. [49] F. Völklein, E. Kessler, Phys. Status Solidi (b) 1987, 143, 121.
[15] B. Abad, D. A. Borca-Tasciuc, M. S. Martin-Gonzalez, Renewable [50] L. Shi, D. Y. Li, C. H. Yu, W. Y. Jang, D. Kim, Z. Yao, P. Kim,
Sustainable Energy Rev. 2017, 76, 1348. A. Majumdar, J. Heat Transfer 2003, 125, 881.
[16] T. M. Tritt, D. Weston, in Thermal Conductivity, Physics of Solids and [51] J. Kim, E. Fleming, Y. Zhou, L. Shi, J. Phys. D: Appl. Phys. 2018, 51,
Liquids (Ed: T. M. Tritt), Springer, Boston, MA 2004, Ch. 2.1, p. 187. 103002.
[17] A. Rudajevova, M. Svantner, D. Vasylyev, O. Musil, V. Lang, Int. J. [52] B. L. Zink, F. Hellman, Solid State Commun. 2004, 129, 199.
Thermophys. 2006, 27, 1241. [53] R. Sultan, A. D. Avery, G. Stiehl, B. L. Zink, J. Appl. Phys. 2009, 105,
[18] J. Sebek, E. Santava, J. Phys.: Conf. Ser. 2009, 150, 012044. 043501.
[19] K. A. Borup, J. de Boor, H. Wang, F. Drymiotis, F. Gascoin, X. Shi, [54] R. Sultan, A. D. Avery, J. M. Underwood, S. J. Mason, D. Bassett,
L. Chen, M. I. Fedorov, E. Müller, B. B. Iversen, G. J. Snyder, Energy B. L. Zink, Phys. Rev. B 2013, 87, 214305.
Environ. Sci. 2015, 8, 423. [55] A. Mavrokefalos, M. T. Pettes, F. Zhou, L. Shi, Rev. Sci. Instrum.
[20] A. Kovalsky, L. Wang, G. T. Marek, C. Burda, J. S. Dyck, J. Phys. 2007, 78, 034901.
Chem. C 2017, 121, 3228. [56] B. L. Zink, A. D. Avery, R. Sultan, D. Bassett, M. R. Pufall, Solid
[21] A. Pisoni, J. Jacimovic, O. S. Barisic, M. Spina, R. Gaal, L. Forro, State Commun. 2010, 150, 514.
E. Horvath, J. Phys. Chem. Lett. 2014, 5, 2488. [57] A. Weathers, K. Bi, M. T. Pettes, L. Shi, Rev. Sci. Instrum. 2013, 84,
[22] V. K. Zaitsev, M. I. Fedorov, E. A. Gurieva, I. S. Eremin, 084903.
P. P. Konstantinov, A. Y. Samunin, M. V. Vedernikov, Phys. Rev. B [58] M. C. Wingert, Z. C. Chen, S. Kwon, J. Xiang, R. Chen, Rev. Sci.
2006, 74, 045207. Instrum. 2012, 83, 024901.
[23] ASTM, ASTM D5470–06, ASTM International, West Consho- [59] A. Weathers, S. Matsushita, M. T. Pettes, M. Goh, K. Akagi, L. Shi,
hocken, PA, USA 2006. J. Appl. Phys. 2013, 114, 194302.
[24] C. Yu, Y. S. Kim, D. Kim, J. C. Grunlan, Nano Lett. 2008, 8, 4428. [60] V. Singh, T. L. Bougher, A. Weathers, Y. Cai, K. Bi, M. T. Pettes,
[25] H. Huang, C. H. Liu, Y. Wu, S. Fan, Adv. Mater. 2005, 17, 1652. S. A. McMenamin, W. Lv, D. P. Resler, T. R. Gattuso, D. H. Altman,
[26] D. Kim, Y. Kim, K. Choi, J. C. Grunlan, C. H. Yu, ACS Nano 2010, K. H. Sandhage, L. Shi, A. Henry, B. A. Cola, Nat. Nanotechnol.
4, 513. 2014, 9, 384.
[27] C. Meng, C. Liu, S. Fan, Adv. Mater. 2010, 22, 535. [61] A. Weathers, Z. U. Khan, R. Brooke, D. Evans, M. T. Pettes,
[28] C. Yu, K. Choi, L. Yin, J. C. Grunlan, ACS Nano 2011, 5, 7885. J. W. Andreasen, X. Crispin, L. Shi, Adv. Mater. 2015, 27, 2101.
[29] G. P. Moriarty, K. Briggs, B. Stevens, C. Yu, J. C. Grunlan, Energy [62] A. D. Avery, B. H. Zhou, J. Lee, E.-S. Lee, E. M. Miller, R. Ihly,
Technol. 2013, 1, 265. D. Wesenberg, K. S. Mistry, S. L. Guillot, B. L. Zink, Y.-H. Kim,
[30] B. M. Zawilski, R. T. Littleton, T. M. Tritt, Rev. Sci. Instrum. 2001, J. L. Blackburn, A. J. Ferguson, Nat. Energy 2016, 1, 16033.
72, 1770. [63] B. A. MacLeod, N. J. Stanton, I. E. Gould, D. Wesenberg, R. Ihly,
[31] C. Wan, X. Gu, F. Dang, T. Itoh, Y. Wang, H. Sasaki, M. Kondo, Z. R. Owczarczyk, K. E. Hurst, C. S. Fewox, C. N. Folmar,
K. Koga, K. Yabuki, G. J. Snyder, R. Yang, K. Koumoto, Nat. Mater. K. Holman Hughes, B. L. Zink, J. L. Blackburn, A. J. Ferguson,
2015, 14, 622. Energy Environ. Sci. 2017, 10, 2168.
[32] H. Imai, Y. Shimakawa, Y. Kubo, Phys. Rev. B 2001, 64. [64] K. Choi, S. L. Kim, S. I. Yi, J. H. Hsu, C. Yu, ACS Appl. Mater. Inter-
[33] T. Varghese, C. Hollar, J. Richardson, N. Kempf, C. Han, faces 2018, 10, 23891.
P. Gamarachchi, D. Estrada, R. J. Mehta, Y. Zhang, Sci. Rep. 2016, [65] H. Ushirokita, H. Tada, Chem. Lett. 2016, 45, 735.
6, 33135. [66] G.-H. Kim, J. Kim, K. P. Pipe, Appl. Phys. Lett. 2016, 108, 093301.
[34] T. M. Tritt, in Thermoelectrics Handbook: Macro to Nano (Ed: [67] A. Greppmair, B. Stoib, N. Saxena, C. Gerstberger, P. Muller-
D. M. Rowe), CRC Press, Boca Raton 2006, Ch. 23, p. 23. Buschbaum, M. Stutzmann, M. S. Brandt, Rev. Sci. Instrum. 2017,
[35] ASTM, ASTM E1225–09, ASTM International, West Consho- 88, 044903.
hocken, PA, USA 2009. [68] N. Saxena, M. Čorić, A. Greppmair, J. Wernecke, M. Pflüger,
[36] ASTM, ASTM E1530–11, ASTM International, West Consho- M. Krumrey, M. S. Brandt, E. M. Herzig, P. Müller-Buschbaum,
hocken, PA, USA 2011. Adv. Electron. Mater. 2017, 3, 1700181.

Adv. Electron. Mater. 2019, 1900167 1900167 (24 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

[69] J. Liu, X. Wang, D. Li, N. E. Coates, R. A. Segalman, D. G. Cahill, [102] J. Liu, J. Zhu, M. Tian, X. Gu, A. Schmidt, R. Yang, Rev. Sci.
Macromolecules 2015, 48, 585. Instrum. 2013, 84, 034902.
[70] Q. Wei, M. Mukaida, K. Kirihara, T. Ishida, ACS Macro Lett. 2014, [103] X. Zheng, D. G. Cahill, J. C. Zhao, Adv. Eng. Mater. 2005, 7, 622.
3, 948. [104] J. P. Feser, J. Liu, D. G. Cahill, Rev. Sci. Instrum. 2014, 85, 104903.
[71] W. J. Parker, R. J. Jenkins, C. P. Butler, G. L. Abbott, J. Appl. Phys. [105] M. Scholdt, H. Do, J. Lang, A. Gall, A. Colsmann, U. Lemmer,
1961, 32, 1679. J. D. Koenig, M. Winkler, H. Boettner, J. Electron. Mater. 2010, 39,
[72] P. Schoderböck, H. Klocker, L. S. Sigl, G. Seeber, Int. J. Thermo- 1589.
phys. 2009, 30, 599. [106] T. Juntunen, H. Jussila, M. Ruoho, S. Liu, G. Hu, T. Albrow-Owen,
[73] V. Khuu, M. Osterman, A. Bar-Cohen, M. Pecht, IEEE Trans. L. W. T. Ng, R. C. T. Howe, T. Hasan, Z. Sun, I. Tittonen, Adv.
Compon., Packag., Manuf. Technol. 2011, 1, 1015. Funct. Mater. 2018, 28, 1800480.
[74] D. Zhao, X. Qian, X. Gu, S. A. Jajja, R. Yang, J. Electron. Packag. [107] G. H. Kim, L. Shao, K. Zhang, K. P. Pipe, Nat. Mater. 2013, 12,
2016, 138, 040802. 719.
[75] H. Yan, N. Sada, N. Toshima, J. Therm. Anal. Calorim. 2002, 69, [108] S. E. Gustafsson, E. Karawacki, M. A. Chohan, J. Phys. D: Appl.
881. Phys. 1986, 19, 727.
[76] W. Q. Ao, L. Wang, J. Q. Li, F. Pan, C. N. Wu, J. Electron. Mater. [109] S. E. Gustafsson, Rev. Sci. Instrum. 1991, 62, 797.
2011, 40, 2027. [110] M. Gustavsson, E. Karawacki, S. E. Gustafsson, Rev. Sci. Instrum.
[77] K. Zhang, Y. Zhang, S. Wang, Sci. Rep. 2013, 3, 3448. 1994, 65, 3856.
[78] W. Zheng, P. Bi, H. Kang, W. Wei, F. Liu, J. Shi, L. Peng, Z. Wang, [111] V. Bohac, M. K. Gustavsson, L. Kubicar, S. E. Gustafsson, Rev. Sci.
R. Xiong, Appl. Phys. Lett. 2014, 105, 023901. Instrum. 2000, 71, 2452.
[79] Y. Liu, Z. Song, Q. Zhang, Z. Zhou, Y. Tang, L. Wang, J. Zhu, [112] Y. He, Thermochim. Acta 2005, 436, 122.
W. Luo, W. Jiang, RSC Adv. 2015, 5, 45106. [113] Y. Li, C. Shi, J. Liu, E. Liu, J. Shao, Z. Chen, D. J. Dorantes-Gonzalez,
[80] Q. Yao, Q. Wang, L. Wang, L. Chen, Energy Environ. Sci. 2014, 7, X. Hu, Meas. Sci. Technol. 2014, 25, 015006.
3801. [114] J. Cha, J. Seo, S. Kim, J. Therm. Anal. Calorim. 2012, 109, 295.
[81] T. Baba, N. Taketoshi, T. Yagi, Jpn. J. Appl. Phys. 2011, 50, 11RA01. [115] Y. Sun, P. Sheng, C. Di, F. Jiao, W. Xu, D. Qiu, D. Zhu, Adv. Mater.
[82] C. A. Paddock, G. L. Eesley, J. Appl. Phys. 1986, 60, 285. 2012, 24, 932.
[83] P. M. Norris, A. P. Caffrey, R. J. Stevens, J. M. Klopf, J. T. McLeskey, [116] J. Wu, Y. Sun, W.-B. Pei, L. Huang, W. Xu, Q. Zhang, Synth. Met.
A. N. Smith, Rev. Sci. Instrum. 2003, 74, 400. 2014, 196, 173.
[84] P. E. Hopkins, J. R. Serrano, L. Phinney, S. P. Kearney, [117] J. Wu, Y. Sun, W. Xu, Q. Zhang, Synth. Met. 2014, 189, 177.
T. W. Grasser, C. T. Harris, J. Heat Transfer 2010, 132, 10. [118] M. Aghelinejad, S. N. Leung, Composites, Part B 2018, 145, 100.
[85] D. G. Cahill, Rev. Sci. Instrum. 2004, 75, 5119. [119] R. Coquard, E. Coment, G. Flasquin, D. Baillis, Int. J. Therm. Sci.
[86] J. Zhu, D. Tang, W. Wang, J. Liu, K. W. Holub, R. Yang, J. Appl. 2013, 65, 242.
Phys. 2010, 108, 094315. [120] S. P. Ashby, J. García-Cañadas, G. Min, Y. Chao, J. Electron. Mater.
[87] O. W. Käding, H. Skurk, K. E. Goodson, Appl. Phys. Lett. 1994, 65, 2013, 42, 1495.
1629. [121] B. Dörling, J. D. Ryan, J. D. Craddock, A. Sorrentino,
[88] D. Chu, M. Touzelbaev, K. E. Goodson, S. Babin, R. F. Pease, A. E. Basaty, A. Gomez, M. Garriga, E. Pereiro, J. E. Anthony,
J. Vac. Sci. Technol., B: Microelectron. Nanometer Struct.–Process., M. C. Weisenberger, A. R. Goñi, C. Müller, M. Campoy-Quiles, Adv.
Meas., Phenom. 2001, 19, 2874. Mater. 2016, 28, 2782.
[89] D. J. Yang, Q. Zhang, G. Chen, S. F. Yoon, J. Ahn, S. G. Wang, [122] D. Kiefer, L. Yu, E. Fransson, A. Gómez, D. Primetzhofer,
Q. Zhou, Q. Wang, J. Q. Li, Phys. Rev. B 2002, 66, 165440. A. Amassian, M. Campoy-Quiles, C. Müller, Adv. Sci. 2017, 4,
[90] M. A. Panzer, G. Zhang, D. Mann, X. Hu, E. Pop, H. Dai, 1600203.
K. E. Goodson, J. Heat Transfer 2008, 130, 052401. [123] R. Kroon, J. D. Ryan, D. Kiefer, L. Yu, J. Hynynen, E. Olsson,
[91] A. Neubrand, J. Dadda, E. Mueller, S. Perlt, T. Höche, J. Electron. C. Müller, Adv. Funct. Mater. 2017, 27, 1704183.
Mater. 2013, 42, 2165. [124] D. Tan, J. Zhao, C. Gao, H. Wang, G. Chen, D. Shi, ACS Appl.
[92] R. J. Stoner, H. J. Maris, Phys. Rev. B 1993, 48, 16373. Mater. Interfaces 2017, 9, 21820.
[93] H. E. Elsayed-Ali, T. B. Norris, M. A. Pessot, G. A. Mourou, Phys. [125] O. Almanza, M. A. Rodriguez-Perez, J. A. De Saja, J. Polym. Sci.,
Rev. Lett. 1987, 58, 1212. Part B: Polym. Phys. 2004, 42, 1226.
[94] R. W. Schoenlein, W. Z. Lin, J. G. Fujimoto, G. L. Eesley, Phys. Rev. [126] J. K. Chen, X. C. Gui, Z. W. Wang, Z. Li, R. Xiang, K. L. Wang,
Lett. 1987, 58, 1680. D. H. Wu, X. G. Xia, Y. F. Zhou, Q. Wang, Z. K. Tang, L. D. Chen,
[95] R. B. Wilson, B. A. Apgar, L. W. Martin, D. G. Cahill, Opt. Express ACS Appl. Mater. Interfaces 2012, 4, 81.
2012, 20, 28829. [127] D. G. Cahill, R. O. Pohl, Phys. Rev. B 1987, 35, 4067.
[96] A. J. Schmidt, X. Chen, G. Chen, Rev. Sci. Instrum. 2008, 79, [128] D. G. Cahill, Rev. Sci. Instrum. 1990, 61, 802.
114902. [129] S. M. Lee, D. G. Cahill, J. Appl. Phys. 1997, 81, 2590.
[97] W. S. Capinski, H. J. Maris, T. Ruf, M. Cardona, K. Ploog, [130] S. M. Lee, D. G. Cahill, R. Venkatasubramanian, Appl. Phys. Lett.
D. S. Katzer, Phys. Rev. B 1999, 59, 8105. 1997, 70, 2957.
[98] J. C. Duda, P. E. Hopkins, Y. Shen, M. C. Gupta, Appl. Phys. Lett. [131] S. T. Huxtable, Ph.D. Thesis, University of California, Berkeley,
2013, 102, 251912. 2002.
[99] D. Beretta, A. J. Barker, I. Maqueira-Albo, A. Calloni, G. Bussetti, [132] T. Borca-Tasciuc, A. R. Kumar, G. Chen, Rev. Sci. Instrum. 2001, 72,
G. Dell’Erba, A. Luzio, L. Duo, A. Petrozza, G. Lanzani, M. Caironi, 2139.
ACS Appl. Mater. Interfaces 2017, 9, 18151. [133] Y. S. Ju, K. Kurabayashi, K. E. Goodson, Thin Solid Films 1999, 339,
[100] A. K. K. Kyaw, T. A. Yemata, X. Wang, S. L. Lim, W. S. Chin, 160.
K. Hippalgaonkar, J. Xu, Macromol. Mater. Eng. 2018, 303, [134] K. C. See, J. P. Feser, C. E. Chen, A. Majumdar, J. J. Urban,
1700429. R. A. Segalman, Nano Lett. 2010, 10, 4664.
[101] K. J. Erickson, F. Leonard, V. Stavila, M. E. Foster, C. D. Spataru, [135] D. Rodin, S. K. Yee, Rev. Sci. Instrum. 2017, 88, 014902.
R. E. Jones, B. M. Foley, P. E. Hopkins, M. D. Allendorf, A. A. Talin, [136] J. Sun, M. L. Yeh, B. J. Jung, B. Zhang, J. Feser, A. Majumdar,
Adv. Mater. 2015, 27, 3453. H. E. Katz, Macromolecules 2010, 43, 2897.

Adv. Electron. Mater. 2019, 1900167 1900167 (25 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

[137] O. Bubnova, Z. U. Khan, A. Malti, S. Braun, M. Fahlman, [168] H. H. Roh, J. S. Lee, D. L. Kim, J. Park, K. Kim, O. Kwon,
M. Berggren, X. Crispin, Nat. Mater. 2011, 10, 429. S. H. Park, Y. K. Choi, A. Majumdar, J. Vac. Sci. Technol., B: Micro-
[138] Y. Yao, M. Shahi, M. M. Payne, J. E. Anthony, J. W. Brill, J. Mater. electron. Nanometer Struct.–Process., Meas., Phenom. 2006, 24,
Chem. C 2016, 4, 8817. 2398.
[139] D. Yuan, D. Huang, C. Zhang, Y. Zou, C. A. Di, X. Zhu, D. Zhu, [169] H. H. Roh, J. S. Lee, D. L. Kim, J. Park, K. Kim, O. Kwon,
ACS Appl. Mater. Interfaces 2017, 9, 28795. S. H. Park, Y. K. Choi, A. Majumdar, J. Vac. Sci. Technol., B: Micro-
[140] D. Huang, H. Yao, Y. Cui, Y. Zou, F. Zhang, C. Wang, H. Shen, electron. Nanometer Struct.–Process., Meas., Phenom. 2006, 24,
W. Jin, J. Zhu, Y. Diao, W. Xu, C. A. Di, D. Zhu, J. Am. Chem. Soc. 2405.
2017, 139, 13013. [170] S. Lefèvre, S. Volz, J.-B. Saulnier, C. Fuentes, N. Trannoy, Rev. Sci.
[141] A. R. Abramson, W. C. Kim, S. T. Huxtable, H. Yan, Y. Wu, Instrum. 2003, 74, 2418.
A. Majumdar, C. L. Tien, P. Yang, J. Microelectromech. Syst. 2004, [171] S. Lefèvre, J. B. Saulnier, C. Fuentes, S. Volz, Superlattices Micro-
13, 505. struct. 2004, 35, 283.
[142] N. E. Coates, S. K. Yee, B. McCulloch, K. C. See, A. Majumdar, [172] S. Lefèvre, S. Volz, Rev. Sci. Instrum. 2005, 76, 033701.
R. A. Segalman, J. J. Urban, Adv. Mater. 2013, 25, 1629. [173] L. David, S. Gomès, M. Raynaud, J. Phys. D: Appl. Phys. 2007, 40,
[143] S. Rausch, D. Rauh, C. Deibel, S. Vidi, H. P. Ebert, Int. J. Thermo- 4337.
phys. 2013, 34, 820. [174] E. Puyoo, S. Grauby, J.-M. Rampnoux, E. Rouvière, S. Dilhaire, Rev.
[144] C. Dames, Annu. Rev. Heat Transfer 2013, 16, 7. Sci. Instrum. 2010, 81, 073701.
[145] Y. Jin, S. Nola, K. P. Pipe, M. Shtein, J. Appl. Phys. 2013, 114, [175] A. Altes, R. Heiderhoff, L. J. Balk, J. Phys. D: Appl. Phys. 2004, 37,
194303. 952.
[146] L. Lu, W. Yi, D. L. Zhang, Rev. Sci. Instrum. 2001, 72, 2996. [176] A. A. Wilson, M. Muñoz Rojo, B. Abad, J. A. Perez, J. Maiz,
[147] W. Yi, L. Lu, D. L. Zhang, Z. W. Pan, S. S. Xie, Phys. Rev. B 1999, J. Schomacker, M. Martín-Gonzalez, D.-A. Borca-Tasciuc,
59, R9015. T. Borca-Tasciuc, Nanoscale 2015, 7, 15404.
[148] T. Y. Choi, D. Poulikakos, J. Tharian, U. Sennhauser, Appl. Phys. [177] M. Chirtoc, J. Gibkes, R. Wernhardt, J. Pelzl, A. Wieck, Rev. Sci.
Lett. 2005, 87, 3. Instrum. 2008, 79, 093703.
[149] J. S. Heron, T. Fournier, N. Mingo, O. Bourgeois, Nano Lett. 2009, [178] E. Puyoo, S. Grauby, J.-M. Rampnoux, E. Rouvière, S. Dilhaire,
9, 1861. J. Appl. Phys. 2011, 109, 024302.
[150] K. M. Lee, T. Y. Choi, S. K. Lee, D. Poulikakos, Nanotechnology [179] S. Grauby, E. Puyoo, J.-M. Rampnoux, E. Rouvière, S. Dilhaire,
2010, 21, 125301. J. Phys. Chem. C 2013, 117, 9025.
[151] G. Li, D. Liang, R. L. J. Qiu, X. P. A. Gao, Appl. Phys. Lett. 2013, [180] M. M. Rojo, S. Grauby, J. M. Rampnoux, O. Caballero-Calero,
102, 043104. M. Martin-Gonzalez, S. Dilhaire, J. Appl. Phys. 2013, 113, 054308.
[152] M. N. Ou, S. R. Harutyunyan, S. J. Lai, C. D. Chen, T. J. Yang, [181] M. M. Rojo, J. Martín, S. Grauby, T. Borca-Tasciuc, S. Dilhaire,
Y. Y. Chen, Phys. Status Solidi (b) 2007, 244, 4512. M. Martin-Gonzalez, Nanoscale 2014, 6, 7858.
[153] S. W. Finefrock, Y. Wang, J. B. Ferguson, J. V. Ward, H. Fang, [182] M. M. Rojo, J. Martín, S. Grauby, T. Borca-Tasciuc, S. Dilhaire,
J. E. Pfluger, D. S. Dudis, X. Ruan, Y. Wu, Nano Lett. 2013, 13, 5006. M. Martin-Gonzalez, Nanoscale 2015, 7, 4256.
[154] K.-Y. Chun, S. H. Kim, M. K. Shin, Y. T. Kim, G. M. Spinks, [183] J. Maiz, M. Muñoz Rojo, B. Abad, A. A. Wilson, A. Nogales,
A. E. Aliev, R. H. Baughman, S. J. Kim, Nanotechnology 2013, 24, D.-A. Borca-Tasciuc, T. Borca-Tasciuc, M. Martín-González, RSC
165401. Adv. 2015, 5, 66687.
[155] Y. Sun, L. Qiu, L. Tang, H. Geng, H. Wang, F. Zhang, D. Huang, [184] B. Abad, I. Alda, P. Diaz-Chao, H. Kawakami, A. Almarza,
W. Xu, P. Yue, Y.-s. Guan, F. Jiao, Y. Sun, D. Tang, C.-a. Di, Y. Yi, D. Amantia, D. Gutierrez, L. Aubouy, M. Martín-González,
D. Zhu, Adv. Mater. 2016, 28, 3351. J. Mater. Chem. A 2013, 1, 10450.
[156] J. Zhang, G. Song, L. Qiu, Y. Feng, J. Chen, J. Yan, L. Liu, X. Huang, [185] J. Bodzenta, J. Juszczyk, M. Chirtoc, Rev. Sci. Instrum. 2013, 84,
Y. Cui, Y. Sun, W. Xu, D. Zhu, Macromol. Rapid Commun. 2018, 39, 093702.
1800283. [186] Y. Zhu, Int. J. Heat Mass Transfer 2016, 92, 784.
[157] F. Huewe, A. Steeger, K. Kostova, L. Burroughs, I. Bauer, [187] A. Prasad, A. Ambirajan, Int. J. Therm. Sci. 2018, 134, 216.
P. Strohriegl, V. Dimitrov, S. Woodward, J. Pflaum, Adv. Mater. [188] A. Green, L. E. J. Cowles, J. Sci. Instrum. 1960, 37, 349.
2017, 29, 1605682. [189] D. J. McNeill, J. Appl. Phys. 1962, 33, 597.
[158] H. Zhang, Y. Yao, M. M. Payne, J. E. Anthony, J. W. Brill, Appl. Phys. [190] R. Homma, Y. Hasegawa, H. Terakado, H. Morita, T. Komine, Jpn.
Lett. 2014, 105, 073302. J. Appl. Phys. 2015, 54, 026602.
[159] A. Sikora, H. Ftouni, J. Richard, C. Hébert, D. Eon, F. Omnès, [191] S. Shin, R. Kumar, J. W. Roh, D. S. Ko, H. S. Kim, S. I. Kim, L. Yin,
O. Bourgeois, Rev. Sci. Instrum. 2012, 83, 054902. S. M. Schlossberg, S. Cui, J. M. You, S. Kwon, J. Zheng, J. Wang,
[160] A. Sikora, H. Ftouni, J. Richard, C. Hébert, D. Eon, F. Omnès, R. Chen, Sci. Rep. 2017, 7, 7317.
O. Bourgeois, Rev. Sci. Instrum. 2013, 84, 029901. [192] T. Kurihara, J. Morikawa, T. Hashimoto, Int. J. Thermophys. 1997,
[161] X. Zhang, C. P. Grigoropoulos, Rev. Sci. Instrum. 1995, 66, 1115. 18, 505.
[162] A. T. Ramu, J. E. Bowers, Rev. Sci. Instrum. 2012, 83, 124903. [193] H. J. Chang, J. Morikawa, T. Hashimoto, J. Appl. Polym. Sci. 2006,
[163] N. Bodenschatz, A. Liemert, S. Schnurr, U. Wiedwald, P. Ziemann, 99, 1104.
Rev. Sci. Instrum. 2013, 84, 084904. [194] J. Morikawa, T. Hashimoto, J. Appl. Phys. 2009, 105, 113506.
[164] H. Wang, F. Yang, Y. Guo, K. Peng, D. Wang, W. Chu, S. Zheng, [195] C. L. Choy, W. P. Leung, Y. K. Ng, J. Polym. Sci., Part B: Polym. Phys.
Measurement 2019, 131, 204. 1987, 25, 1779.
[165] D. G. Cahill, W. K. Ford, K. E. Goodson, G. D. Mahan, [196] Y. Hiroshige, M. Ookawa, N. Toshima, Synth. Met. 2007, 157, 467.
A. Majumdar, H. J. Maris, R. Merlin, S. R. Phillpot, J. Appl. Phys. [197] L. Zhang, T. Goto, I. Imae, Y. Sakurai, Y. Harima, J. Polym. Sci.,
2003, 93, 793. Part B: Polym. Phys. 2017, 55, 524.
[166] S. Gomès, A. Assy, P.-O. Chapuis, Phys. Status Solidi A 2015, 212, 477. [198] K. Suemori, T. Kamata, Synth. Met. 2017, 227, 177.
[167] G. Wielgoszewski, T. Gotszalk, in Advances in Imaging and Electron [199] A. J. Schmidt, R. Cheaito, M. Chiesa, Rev. Sci. Instrum. 2009, 80,
Physics, Vol. 190 (Ed: P. W. Hawkes), Academic Press, Elsevier Ltd., 094901.
Waltham 2015, Ch. 4, p. 177. [200] A. J. Schmidt, R. Cheaito, M. Chiesa, J. Appl. Phys. 2010, 107, 5.

Adv. Electron. Mater. 2019, 1900167 1900167 (26 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
www.advelectronicmat.de

[201] W.-L. Ong, E. S. O’Brien, P. S. Dougherty, D. W. Paley, C. Fred [222] K. Horne, H. Ban, A. Mandelis, A. Matvienko, Mater. Sci. Eng., B
Higgs III, A. J. McGaughey, J. A. Malen, X. Roy, Nat. Mater. 2017, 2012, 177, 164.
16, 83. [223] A. Wolf, P. Pohl, R. Brendel, J. Appl. Phys. 2004, 96, 6306.
[202] W.-L. Ong, S. M. Rupich, D. V. Talapin, A. J. H. McGaughey, [224] A. Mendioroz, R. Fuente-Dacal, E. Apinaniz, A. Salazar, Rev. Sci.
J. A. Malen, Nat. Mater. 2013, 12, 410. Instrum. 2009, 80, 074904.
[203] J. A. Malen, K. Baheti, T. Tong, Y. Zhao, J. A. Hudgings, [225] L. I. Giri, S. Tuli, M. Sharma, P. Bugnon, H. Berger, A. Magrez,
A. Majumdar, J. Heat Transfer 2011, 133, 081601. Mater. Lett. 2014, 115, 106.
[204] I. Hatta, Y. Sasuga, R. Kato, A. Maesono, Rev. Sci. Instrum. 1985, [226] M. Streza, S. Longuemart, E. Guilmeau, K. Strzalkowski, K. Touati,
56, 1643. M. Depriester, A. Maignan, A. Hadj Sahraoui, J. Phys. D: Appl.
[205] R. Kato, A. Maesono, R. P. Tye, Int. J. Thermophys. 2001, 22, Phys. 2016, 49, 285601.
617. [227] R. Islam, R. Chan-Yu-King, J. F. Brun, C. Gors, A. Addad,
[206] X. Y. Yu, L. Zhang, G. Chen, Rev. Sci. Instrum. 1996, 67, 2312. M. Depriester, A. Hadj-Sahraoui, F. Roussel, Nanotechnology 2014,
[207] R. Kato, A. Maesono, I. Hatta, Jpn. J. Appl. Phys. 1993, 32, 3656. 25, 475705.
[208] H. Kato, K. Nara, M. Okaji, Cryogenics 2001, 41, 373. [228] F. Roussel, R. Chen Yu King, M. Kuriakose, M. Depriester,
[209] D. A. Borca-Tasciuc, G. Chen, J. Appl. Phys. 2005, 97, 084303. A. Hadj-Sahraoui, C. Gors, A. Addad, J.-F. Brun, Synth. Met. 2015,
[210] T. Borca-Tasciuc, D. A. Borca-Tasciuc, G. Chen, IEEE Trans. 199, 196.
Compon. Packag. Technol. 2007, 30, 609. [229] J. W. Brill, M. Shahi, M. M. Payne, J. Edberg, Y. Yao, X. Crispin,
[211] H. Zhang, J. W. Brill, J. Appl. Phys. 2013, 114, 043508. J. E. Anthony, J. Appl. Phys. 2015, 118, 235501.
[212] D. A. Borca-Tasciuc, G. Chen, A. Prieto, M. S. Martín-González, [230] A. Rosencwaig, A. Gersho, J. Appl. Phys. 1976, 47, 64.
A. Stacy, T. Sands, M. A. Ryan, J. P. Fleurial, Appl. Phys. Lett. 2004, [231] F. A. McDonald, G. C. Wetsel, J. Appl. Phys. 1978, 49, 2313.
85, 6001. [232] H. Vargas, L. C. M. Miranda, Phys. Rep. 1988, 161, 43.
[213] B. A. Lunn, J. Unsworth, N. G. Booth, P. C. Innis, J. Mater. Sci. [233] H. Hu, X. Wang, X. Xu, J. Appl. Phys. 1999, 86, 3953.
1993, 28, 5092. [234] O. Pessoa, C. L. Cesar, N. A. Patel, H. Vargas, C. C. Ghizoni,
[214] A. Malti, J. Edberg, H. Granberg, Z. U. Khan, J. W. Andreasen, L. C. M. Miranda, J. Appl. Phys. 1986, 59, 1316.
X. Liu, D. Zhao, H. Zhang, Y. Yao, J. W. Brill, I. Engquist, [235] J. A. Balderas-López, Rev. Sci. Instrum. 2003, 74, 1338.
M. Fahlman, L. Wagberg, X. Crispin, M. Berggren, Adv. Sci. 2016, [236] J. A. Balderas-López, Rev. Sci. Instrum. 2006, 77, 064902.
3, 1500305. [237] A. Lachaine, P. Poulet, Appl. Phys. Lett. 1984, 45, 953.
[215] F. Jiao, A. Naderi, D. Zhao, J. Schlueter, M. Shahi, J. Sundström, [238] X. Wang, H. Hu, X. Xu, J. Heat Transfer 2001, 123, 138.
H. Granberg, J. Edberg, U. Ail, J. Brill, T. Lindström, M. Berggren, [239] B. A. Cola, R. Karru, C. Cheng, X. Xu, T. S. Fisher, IEEE Trans.
X. Crispin, J. Mater. Chem. A 2017, 5, 16883. Compon. Packag. Technol. 2008, 31, 46.
[216] R. Santos, L. C. M. Miranda, J. Appl. Phys. 1981, 52, 4194. [240] J. A. Balderas-López, A. Mandelis, Rev. Sci. Instrum. 2003, 74, 5219.
[217] L. Fabbri, P. Fenici, Rev. Sci. Instrum. 1995, 66, 3593. [241] K. G. Biswas, T. D. Sands, B. A. Cola, X. Xu, Appl. Phys. Lett. 2009,
[218] M. Depriester, P. Hus, S. Delenclos, A. H. Sahraoui, Rev. Sci. 94, 223116.
Instrum. 2005, 76, 074902. [242] T. A. El-Brolossy, S. S. Ibrahim, Phys. Scr. 2014, 89, 105701.
[219] R. Fuente, E. Apinaniz, A. Mendioroz, A. Salazar, J. Appl. Phys. [243] M. K. Smith, V. Singh, K. Kalaitzidou, B. A. Cola, ACS Nano 2015,
2011, 110, 033515. 9, 1080.
[220] P. Martinez-Torres, A. Mandelis, J. J. Alvarado-Gil, J. Appl. Phys. [244] M. Kuriakose, M. Depriester, R. Chan Yu King, F. Roussel, A. Hadj
2009, 106, 114906. Sahraoui, J. Appl. Phys. 2013, 113, 044502.
[221] J. A. Garcia, A. Mandelis, B. Farahbakhsh, C. Lebowitz, I. Harris, [245] K. Touati, M. Depriester, M. Kuriakose, A. H. Sahraoui, Rev. Sci.
Int. J. Thermophys. 1999, 20, 1587. Instrum. 2015, 86, 094901.

Adv. Electron. Mater. 2019, 1900167 1900167 (27 of 27) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

You might also like