You are on page 1of 8

Microwave microscopy of diamond semiconductor structures

A. N. Reznik, S. A. Korolyov, and M. N. Drozdov

Citation: Journal of Applied Physics 121, 164503 (2017); doi: 10.1063/1.4982676


View online: http://dx.doi.org/10.1063/1.4982676
View Table of Contents: http://aip.scitation.org/toc/jap/121/16
Published by the American Institute of Physics
JOURNAL OF APPLIED PHYSICS 121, 164503 (2017)

Microwave microscopy of diamond semiconductor structures


A. N. Reznik,a) S. A. Korolyov, and M. N. Drozdov
Institute for Physics of Microstructures of the Russian Academy of Sciences, GSP-105,
Nizhny Novgorod 603950, Russia
(Received 16 February 2017; accepted 16 April 2017; published online 28 April 2017)
A scanning near-field microwave microscope (SNMM) was used to study resistance Rsh of a boron
delta-doped epilayer of diamond grown on a high pressure and high temperature substrate.
Measurements were performed with an 1.4 GHz working frequency and an 85 lm space
resolution microscope on samples with a lateral dimension of 3  3 mm2. Some substrates featured
a crystalline structure defect over which the epilayer resistance Rsh was seen to increase by more
than an order of 1 mm linear dimensions of the high-ohmic region. The SNMM measurement
data revealed some substrates to have nonuniform conductivity, i.e., a high-ohmic area in the cen-
tral part surrounded by a conducting edge region. In the latter case, the SNMM method allowed
determining a surface distribution of epilayer resistance Rsh, undistorted by the shunting influence
of the substrate. The reliability of the SNMM results is confirmed by the local four-probe resistance
measurements. At the same time, the alternative methods such as the non-local van der Pauw
method and secondary ion mass spectroscopy failed to detect the nonuniformity of the structure
conducting properties, established by SNMM. The obtained values for a hole concentration of
Np  1.7  1020 cm3 and a mobility of lH  15 cm2/(V s) are assumed to possibly pertain to the
diamond delta-layer. Published by AIP Publishing. [http://dx.doi.org/10.1063/1.4982676]

INTRODUCTION material. In the past 15–20 years, considerable success has


been achieved in quantitative SNMM diagnostics,9–11,20–25
Present-day manufacturing technologies for devices and
making it possible to demonstrate the characterization poten-
elements of semiconductor micro- and nanoelectronics
tialities for semiconductors of yet a fairly simple structure—
involve characterization of samples in a variety of methods
bulk homogeneous or thin film ones.6,16,20,24,26 These
and techniques. Among the most widely used ones are sec-
achievements give reason to believe that SNMM can find
ondary ion mass spectroscopy (SIMS), capacitance versus
wider applications in semiconductor technologies.
voltage C(V) measurements, Hall effects combined with four
The key direction in the SNMM development over the
probe resistivity measurements (commonly referred to as
the van der Pauw (VDP) method), optical ellipsometry, elec- recent years has been towards achieving extremely high
tron microscopy, infrared Fourier spectroscopy, probe (nanometer and subnanometer) spatial resolution, being
microscopy, X-ray structural analysis, and many others. called for by the corresponding size of today’s electronic
Simultaneous application of several probing techniques components. It seems that there is an appropriate area of
allows one to get more information on structures in question. application for medium-resolution SNMM as well, in particu-
Moreover, to upgrade the data reliability, it is useful to lar, in working-out manufacturing technologies for new semi-
employ tools that will ensure the determination of similar conductor materials. One promising material under active
parameters (for example, conducting or structural properties) investigation is semiconductor diamond.27 It is expected that
through independent measurements. In this connection of the predicted outstanding properties of diamond (large carrier
great potential value are non-traditional methods that are mobility, high breakdown voltage, and exceptional thermal
currently under development and testing. Of these, a very conductivity) will help improve the characteristics of semi-
promising technique is scanning near-field microwave conductor devices for high power and high frequency elec-
microscopy (SNMM).1–3 This method is characterized by a tronics. The main problems of semiconductor diamond
high resolving power that exceeds the working wavelength fabrication technology now include producing large-size
by many orders of magnitude. Modern microscopes in the (>25 mm) single crystal diamond substrates, attaining high
centimeter and decimeter wave range are capable of nanome- values of free-carrier concentration (>5  1020 cm3) and
ter resolution4–7 down to the atomic scale.8 Yet, a great mobility (>103 cm2/(V s)), forming ultra-thin conducting
amount of research so far has been done based on medium layers (<2 nm) by delta-doping, and realizing surface-
resolution (1–103 lm) devices.9–20 The important advantages homogeneous characteristics of structures. The current
of SNMM rest on its capability for non-invasive nondestruc- single-crystal diamond substrates are not more than 15 mm in
tive probing of objects not only on the surface but also in size. Due to the high cost of the substrates, studies are usually
depth. Another essential characteristic is the high sensitivity conducted on samples of much smaller lateral dimensions
of the microscope to the conducting properties of the (the standard size is 3  3 mm2), which is quite an obstacle
for microwave diagnostics. Of the widest use are the sub-
a)
Electronic mail: reznik@ipm.sci-nnov.ru strates grown by the temperature gradient method at high

0021-8979/2017/121(16)/164503/7/$30.00 121, 164503-1 Published by AIP Publishing.


164503-2 Reznik, Korolyov, and Drozdov J. Appl. Phys. 121, 164503 (2017)

pressure and high temperature (HPHT).28,29 The HPHT sub- change in the Q-factor when the probe comes in contact with
strates, depending on the growth parameters, differ in their the sample. Resistance Rsh is calculated within the model of
crystalline quality and impurity content; they may have an SNMM probe proposed in Ref. 20, which is basically a
growth sectors30 and other structural defects. Substrate monopole antenna. The fitting parameters for the model are
defects reveal about the structural and electrical properties of sought using a series of calibration etalons—dielectric quartz
epilayers that grow on them.31–33 Therefore, epitaxial growth (EQ), sapphire (ES), and conductive silicon plates varying in
has to be controlled by local measurements. In this paper, we a doping degree. For the studied diamond structures, it was
show that a medium-resolution SNMM can serve as an effec- enough to use the two highest-resistance Si plates with resis-
tive tool for quantitative diagnostics of such structures. tivity q ¼ 6000 and 52 X cm (etalons E0 and E1). The plates’
thickness was 0.4–0.7 mm. The microscope resolution power
SAMPLES AND METHODS depends on diameter dt of the probe tip end; according to the
micrography data, dt  40 lm. The actual resolution was
Diamond structures were grown on [100]-oriented
found from measurements of the parameters f and Q by scan-
HPHT substrates with a lateral size of 3  3 mm2. We stud-
ning a sample surface on which the metal and dielectric
ied samples with boron p-doped delta-layers of 2–4 nm
regions were divided by a sharp line. The obtained value for
thickness. The delta-doping of diamond proceeded during
resolution d  85 lm proved to be poorer than expected from
epitaxial growth in a microwave plasma-enhanced chemical
dt measurements. In a study of diamond samples, the probe
vapor deposition (MPCVD) reactor with a rapid gas switch-
tip was scanned over the sample surface {x, y} along the
ing system.34 A typical thickness of the epilayers was axis [ (at y ¼ const) with a step D[  d. In every point of
100 nm, although some samples had a 10–20 nm epilayer. the surface, the tip was brought in a “soft” contact with the
Diborane (B2H6 gas) was used as a source of boron. sample and then moved off the surface to a fixed height
Along with the SNMM, we also used the standard techni- Dz ¼ 0.3 lm, as described in Ref. 20. We measured the
ques for characterization of semiconductor structures, such as microscope resonance curve (frequency dependence of reso-
the SIMS method and VDP method. The SIMS method nator reflectivity), upon which the probe was withdrawn to a
was used for measuring a depth profile of boron atom concen- certain height z, moved one step D[ along the surface,
tration NB(z) on a TOF.SIMS-5 setup. A series of test and the positioning/measuring procedure was repeated.
structures—diamond layers implanted with 50 keV boronions According to the test measurement data from Ref. 20 and
11
B in a varying dose—were fabricated for boron sensitivity some additional studies, the error of Rsh measurements with
calibration. The VDP method35,36 yields hole concentration our coaxial SNMM does not exceed 50% at Rsh > 1 kX/sq.
Np and mobility lH within the epitaxial film-substrate system. Unlike in Ref. 20, we used the method developed therein
In this study, we used two geometries of VDP measurements, to determine resistance Rsh of the epilayer on a conducting
specifically, the sample geometry (VDP-S) and the local substrate and to measure resistivity q of the substrate itself.
lamella-type geometry (VDP-L). For VDP-S measurements, The generalization of the method20 to these situations does not
four ohmic contacts were formed in the corners of a test sam- cause any difficulties. It is suitable for testing planar structures
ple. In the VDP-L technique, five lamellar specimens were of any complexity, with one parameter to be determined Rsh or
shaped on a sample surface as electrically isolated Hall q. The complex permittivity and thickness of all other layers in
crosses, i.e., outside the crosses, the epilayer was removed. the structure must be known. The SNMM model underlying
Each cross had four ohmic contacts formed in its corners (see this technique allows the calculation of informative parameters
the inset in Fig. 5(a)). The overall size of a cross was f and Q (i.e., a solution of the direct problem) for a structure
0.45  0.45 mm2, but the characterization target was a with any finite number of layers. To determine the single
0.15  0.15 mm2 area in the center of the cross. The arrange- unknown parameter of the structure (solution of the inverse
ment of crosses on a sample surface is shown in the insets in problem), we have designed a special computer algorithm,
Figs. 3(b), 5(a), and 5(b). Unlike the SNMM, the VDP-L enabling the partial automation of the measurement technique.
method involves several technological operations to be per- It is obvious that our approach assumes homogeneity of the
formed during specimen fabrication, and the epitaxial film test structure on scale d, i.e., the characteristic scale of lateral
has to be etched off almost completely. Also, note that VDP- inhomogeneity of parameters Rsh and q must be larger than d.
S measurements caused the destruction of the doped film in As a first example of SNMM measurements, Fig. 1 repre-
the region where conducting contacts were formed with linear sents dependencies of resonance frequency f and the Q factor
dimensions of 200–300 lm. The same is true for SIMS on coordinate x along scan lines 1–3 for sample De1. It is a
measurements whose by-effect on a sample is the formation structure with an epilayer of a thickness of 80 nm including
of a 100 lm diameter crater on its surface. five delta-layers varying in a peak concentration of boron
Our microwave microscope design is described in Ref. atoms. A SIMS depth profile of concentration NB(z) is shown
20, where we also provide detailed information on the tech- in Fig. 4(a). Profiles f(x) and Q(x) in Figs. 1(a) and 1(b),
nique for determining resistance Rsh of a semiconductor film respectively, characterize the inhomogeneity of structure De1.
on a dielectric substrate. The microscope is essentially a Higher values for f and Q correlate with higher resistance val-
1.4 GHz coaxial k/2 resonator with a probe in the form of a ues. The inset in Fig. 1(a) (and insets in Figs. 3(a), 3(b), and
sharpened tip, connected to the end of the resonator central 5(a)) shows the high-ohmic region schematically designated
conductor. The informative parameters used to determine the by dark ovals based on the SNMM scan data. The experimen-
value of Rsh are a shift of resonance frequency f and a tal data in Fig. 1(c) are presented in variables n and g, where
164503-3 Reznik, Korolyov, and Drozdov J. Appl. Phys. 121, 164503 (2017)

we can quickly estimate the lateral inhomogeneity of the sam-


ple under investigation. In particular, our experience in
SNMM measurements prior to the Rsh calculation allows us to
assert that sample De1 at points A1 and A3 differs in the resis-
tance value by about an order of magnitude. Besides, in the
low-ohmic region of the epilayer (vicinity of point A3), we
have Rsh > 10 kX/sq. Point Ds0 in Fig. 1(c) is the measure-
ment taken on a dielectric substrate. The location of this point
is determined by the real part of the diamond permittivity e:
e0 ¼ Re(e) ¼ 5.7 (for quartz, sapphire, and undoped silicon,
e0 ¼ 4.5, 10, and 11.7, respectively).
Diamond substrates often contain macroscopic defects of
a crystal structure. Some examples are substrates Ds1 and Ds2
whose 3D surface image is shown in Fig. 2. In contrast to
structures De1, De2, and De3 with an epilayer, we marked the
bare diamond substrates for corresponding samples as Ds1,
Ds2, and Ds3. The substrate images were taken with an
Optical Surface Profiler Zygo NewView 7300. Prior to imag-
ing, the surface had been treated by the H2/O2 plasma. In
Fig. 2, one can see the defective areas of 0.5–1 mm on the
substrates. The observed high resistance inhomogeneity in the
epilayer of sample De1 (see inset in Fig. 1(a)) is, in all proba-
bility, related to the structural defect on the substrate Ds1 sur-
face. The image of the latter in Fig. 2(a) coincides with the
high-ohmic region in the epilayer as measured by SNMM.

RESULTS AND DISCUSSION


Based on the SNMM studies, it has been established that
many of the diamond substrates have a laterally inhomoge-
neous conductivity. Application of the method20 to a bulk

FIG. 1. (a) Resonance frequency f and (b) Q-factor obtained by SNMM


scanning of sample De1 along lines 1–3. (c) Parameters f and Q in normal-
ized variables n and g. A1–A3 are the values for n and g in points on the sur-
face of sample De1, marked in (a) and (b). The line is the calibration curve
gc(n), calculated from measurements on etalons EQ, ES, E0, and E1. Point
Ds0 is the measurement on the dielectric diamond substrate. The inset in (a)
shows scanning lines 1–3 and the approximate location of the high-ohmic
region (dark ovals).

n ¼ (fEQ  f)/fEQ, g ¼ 0.5(1/Q  1/QEQ) are the normalized


resonance frequency and inverse value of the Q-factor; fEQ
and QEQ are parameters f and Q from a contact of the probe
with etalon EQ. The values A1–A3 in Fig. 1(c) correspond to
the three characteristic points on sample De1, indicated in
Figs. 1(a) and 1(b). The calibration curve gc(n) is plotted from
measurements on etalons EQ, ES, E0, and E1. The data in FIG. 2. Surface image of substrates (a) Ds1 and (b) Ds2. 1–3—SNMM scan
Fig. 1(c) provide the basis for realizing a method20 of deter- lines and ovals—the high-ohmic region. The arrow indicates the defect
mining resistance Rsh. Moreover, based on this information, location.
164503-4 Reznik, Korolyov, and Drozdov J. Appl. Phys. 121, 164503 (2017)

sample such as a substrate, as mentioned above, yields resis- SNMM method (for Ds2 and Ds3; see Fig. 3). After the
tivity q. Figures 3(a) and 3(b) give the q profiles of substrates growth of the delta-doped epilayer, the film-substrate system
Ds2 and Ds3. The nonuniform distributions of q(x,y) in Fig. was scanned by SNMM approximately along the same lines.
3—a high-ohmic region with q 10 kX cm in the central The effect of the substrate conductivity was taken into
part, with resistance decreasing towards the edges down to account in calculations of the epilayer resistance Rsh by the
q  0.1–1 kX cm—are quite typical for substrates under study. method described in Ref. 20. Figure 5 shows the Rsh profiles
The high-ohmic area on substrate Ds2 roughly corresponds to of the epitaxial film for samples De2 and De3. Scan lines 1
the location of a defect, as shown in Fig. 2(b), whereas no and 3 on sample De3 are interrupted due to the ohmic con-
macro-defects were observed for Ds3. It should be noted that tacts that had been formed in the corners of the structure for
our VDP-S measurements also revealed appreciable conduc- VDP-S measurements. SIMS depth profiles of boron concen-
tivity in a number of substrates. For example, the conductivity tration NB(z) in samples De2 and De3 are shown in Figs.
of substrate Ds2, obtained by the VDP-S technique, was 4(b) and 4(c), respectively.
q ¼ 0.8 kX cm. A comparison of this value with the SNMM The effect of substrate conductivity is the strongest on
data in Fig. 3(a) clearly shows that such a value of q is charac- SNMM measurements of structures with a high-ohmic epi-
teristic only of a substrate edge region. This conclusion is layer, such as in sample De3 (see Fig. 5(b)). The average
fairly reasonable for objects with inhomogeneous conductivity over-all-points’ value of epilayer resistance is Rsh ¼ 250 kX/
since by nonlocal VDP-S measurements, a direct current is
concentrated in low-resistance areas, which in our case corre-
sponds to the near-edge region of substrate Ds2. In contrast,
local SNMM measurements give a distribution q(x,y) in the
surface plane of a test sample.
The developed technique allowed us to obtain scan pro-
files of resistance Rsh of an epilayer that is grown, in particu-
lar, on an inhomogeneously conducting substrate. The
distribution q(x,y) of the substrate was first measured by the

FIG. 3. Resistivity profiles for substrates (a) Ds2 and (b) Ds3, scanned along
lines 1–3. The insets show scanning lines 1–3 and the approximate location FIG. 4. Depth profiles of boron atom concentration NB(z) for samples (a)
of the high-ohmic regions (dark ovals). The inset in (b) also shows the posi- De1, (b) De2, and (c) De3. The inset in (b) shows points 1–3 on the surface
tion of Hall crosses. of sample De2, at which we obtained profiles NB(z).
164503-5 Reznik, Korolyov, and Drozdov J. Appl. Phys. 121, 164503 (2017)

sq, and root-mean-square deviation rR ¼ 100 kX/sq. Based for De3 are close to the limit. The above estimate of the cut-
on the measured profiles f(x) and Q(x), one could expect a off value, obtained from the fluctuating sensitivity threshold
much higher inhomogeneity of the epilayer. The correspond- of our SNMM with respect to parameters f and Q, is suitable
ing effect was found to be related to the inhomogeneous con- for testing a conducting film deposited on any dielectric sub-
ductivity of the substrate, given a lower resistance of the strate. At Rsh > 500 kX/sq, parameters f and Q measured for
sample edge area. Indeed, a substrate with a thickness of the film-substrate system do not differ, within a fluctuation
0.3 mm and resistivity q ¼ 0.2–11 kX cm (see Fig. 3(b)) has error, from similar parameters for a bare dielectric substrate.
a dc resistance of 7–370 kX/sq. Unlike the dc case, an ac For a bulk homogeneous conducting sample, such as the sub-
field does not penetrate the entire thickness of the substrate. strates under study here, a similar estimation of cutoff resis-
The depth of a microwave field penetration is determined by tivity yields qc  15 kX cm. At q > qc, the SNMM perceives
a probe aperture size dt, and by our estimation, it is about the substrate as a dielectric (point Ds0 in Fig. 1(c)). The
100 lm. So, the effective shunting resistance of a substrate, limit of measurable characteristics can be increased through
arising from SNMM measurements of sample De3, proves to more rigorous temperature/vibration stabilization and other
be 20–1000 kX/sq depending on a position of the measured improvements of SNMM. In particular, modern nanometer-
point. Considering the above values of Rsh for the epilayer, resolution devices6,7 have a higher sensitivity threshold since
the low-ohmic part of the substrate largely contributes to the they are designed to measure significantly smaller variations
SNMM measured dependences f(x,y) and Q(x,y) for the edge of the SNMM resonance curve than mid-resolution micro-
area of sample De3. The developed SNMM method yields a scopes. Naturally, these devices are much more expensive
resistance value for the epilayer that is not distorted by the and harder to be operated. The measurement capabilities of
substrate. According to the data in Fig. 5(b), the epilayer of our SNMM proved fairly adequate to serve the purpose of
sample De3 is, apparently, sufficiently homogeneous, and the this work.
observed spread in values is most likely related to an error in The findings for sample De2 give evidence of the fact
the SNMM measurements of high-ohmic structures with that defects in a substrate crystal structure influence the elec-
Rsh > 100 kX/sq. By estimates, the limit of Rsh measurement trical properties of an epilayer. SNMM scanning results in
for our SNMM is 500 kX/sq, i.e., the Rsh values obtained Fig. 5(a) show the inhomogeneity of epilayer resistance Rsh.
The high-ohmic region of the layer coincides with the area
of structural inhomogeneity on the substrate Ds2 (see inset
in Fig. 4(b)). Note that in this case, the boron concentration
profile NB(z), measured by SIMS in three points on the sam-
ple both in the defect region and beyond, proved to be practi-
cally the same (Fig. 4(b)). So, a substrate crystal structure
defect does not affect the impurity content while reducing
either a concentration or mobility of charge carriers in the
corresponding region of the epilayer. The VDP-S measure-
ment yields resistance Rsh ¼ 8 kX/sq, which characterizes
only the near-edge area of the epitaxial film and is in accord
with the corresponding SNMM data in Fig. 5(a), obtained for
this region.
Rather, low resistance values obtained from the edge
region of the epitaxial film in sample De2, Rsh  7–8 kX/sq,
can be related to a highly doped delta-layer. The fact that
SNMM and VDP-S measurements yielded close Rsh results
for this region means that the VDP-S data can be used for
the estimation of electrophysical parameters in it. Taking the
delta-layer thickness t ¼ 3.5 nm (see SIMS data in Fig. 4(b)),
from the VDP-S data, we found the following characteristics
for the delta-layer: resistivity, q ¼ Rsht  2.5  103 X cm;
hole concentration, Np  1.7  1020 cm3; and mobility,
lH  15 cm2/(V s). With the peak boron concentration being
NB  3  1020 cm3 (see Fig. 4(b)), the dopant ionization
was estimated at 57%. The obtained characteristics do not
contradict the available literature data on the conductivity of
a delta-layer in diamond.37–43 However, an unambiguous
conclusion that these characteristics pertain to the delta-layer
FIG. 5. Scan profiles of the epilayer resistance for samples (a) De2 and (b) can be made only based on further research, as there may be
De3. Right-hand insets of (a) and (b) show scanning lines 1–3 and the posi-
other conductivity channels in the epilayers of diamond, for
tion of crosses I–V under VDP-L measurements. In the right-hand inset of
(a), the dark oval indicates a high-ohmic region. The left-hand inset of (a) example, those arising through the addition of electrically
shows a Hall cross structure. active hydrogen on interfaces.44–46
164503-6 Reznik, Korolyov, and Drozdov J. Appl. Phys. 121, 164503 (2017)

TABLE I. VDP-L measurements of electrical parameters for sample De3 TABLE II. VDP-L measurements of Rsh for sample De2.
and substrate Ds3.
Number of cross Rsh, kX/sq
Number of Rsh, kX/sq lH, cm2/(V s) Ns, 1/cm2
cross (sample/substrate) (sample) (sample) I 54
II 310
I 17/17 1180 3.2  1011 III 4500
II 5/4 680 1.5  1012 IV 47
III 130/- 18 2.8  1012 V Cross damaged
IV 330/- … …
V 60/- 31 3.4  1012
substrate. So, we see that non-local VDP studies of an epilayer
in substrates with inhomogeneous conductivity may not be
Thus, SNMM measurements yielded information on the informative enough. SNMM measurements in the same condi-
epilayer, which is undistorted by the shunting effect and tions yield objective information on a layer resistance.
inhomogeneous conductivity of the substrate. This allowed The results of the VDP-L resistance measurements on
us to study the electrophysical properties of the epilayer and sample De2 are presented in Table II. The position of Hall
delta-layer with a higher accuracy. We were also able to cor- crosses I-V is shown in the inset in Fig. 5(a). The conclu-
rectly use the data from measurements performed by alterna- sions made from the SNMM studies were also found to be
tive methods. In particular, comparing the photograph of true for sample De2: the maximum resistance is observed at
substrate Ds2 in Fig. 2(b) with the results of SNMM scan- crosses II and III over the defect region of the substrate; out-
ning of sample De2 in Fig. 5(a) has led us to a conclusion side this region (crosses I and IV), the epitaxial layer resis-
that the high-ohmic region of the epilayer is most likely due tance decreases. The obtained SNMM and VDP-L values
to a substrate defect. On the other hand, relating the VDP-S of Rsh show a mutually satisfactory agreement (compare Fig.
measurement data to the low-ohmic near-edge region of
5(a) with the data in Table II). The only exception is point
sample De2, based on the SNMM results, allowed us to, a
III at which the difference in the measurement result by an
certain degree of reliability, estimate the electrical parame-
order of magnitude is apparently connected with an error in
ters of the delta-layer.
the SNMM and VDP-L techniques when measuring high val-
The SNMM results are confirmed by the VDP-L meas-
ues of resistance Rsh.
urements. In Table I, we provide the measured values for the
layer resistance Rsh, mobility lH, and sheet density Ns of free CONCLUSION
carriers in structure De3 and its substrate Ds3. The position of
crosses I-V is shown in the insets in Figs. 3(b) and 5(b). The In this paper, it has been demonstrated that the devel-
obtained values for structure De3 are determined by the con- oped SNMM technique is an effective tool for quantitative
ductivity of both the epilayer and the substrate. To identify the characterization of semiconductor doped structures. The
contributions from either of these two conductivity channels, major advantages of our mid-resolution SNMM include local
we measured the substrate resistance after the epitaxial layer and nondestructive non-contact diagnostics that show up
had been fully removed. It should be noted that VDP-L meas- most vividly in the process of developing the technologies
urements on structures with a conducting substrate yield for the fabrication of new materials. One example of such a
approximate data. The use of the VDP technique requires that material is semiconductor diamond. We have studied boron
ohmic contacts be formed at the edges of a test structure, delta-doped CVD diamond structures epitaxially grown on
which was not the case in our VDP-L measurements of sam- an HPHT substrate. Working out a technology for doping
ples with a high conductive substrate. The VDP-L studies con- and characterization of diamond structures involves great
firmed the inhomogeneity of conductivity in substrate Ds3. Its difficulties related, in particular, to small sizes, macroscopic
high-ohmic region is located in the vicinity of crosses III-V, defects, and laterally inhomogeneous conductivity of sub-
where measurements failed due to a high resistance between strates. Thus, the VDP-S method proved to be least informa-
the contacts of the cross. The results Rsh ¼ 60–330 kX/sq for tive in the case when the substrate shunted the doped
sample De3 with an epilayer, which were obtained for crosses epilayer due to high conductivity in the edge region. VDP-L
III-V, are in agreement with the SNMM data (see Fig. 5(b)), measurements allowed us to obtain local data only at some
i.e., they refer to the layer. Resistance Rsh ¼ 4–17 kX/sq, as points in the central part of a sample surface, provided that
well as concentration Ns and mobility lH, measured on the epilayer on the rest of the substrate surface had to be
crosses I and II, characterizes only the substrate that shunts a removed. The SIMS technique yielded high homogeneity of
higher resistance of the epilayer. The same conclusion is the doping impurity concentration when the conducting
made based on the SNMM data given in Figs. 3(b) and 5(b). properties of a structure were strongly inhomogeneous. In
Knowing the substrate thickness and the sheet density Ns, these conditions, when the characterization of diamond sam-
measured using Hall crosses I and II, we can estimate a bulk ples proved to be rather complicated, the SNMM merits
concentration of free carriers in the substrate conducting area, showed up most vividly. Our SNMM method enabled the
which is N  1013–1014 1/cm3. The VDP-S resistance mea- mapping of resistance Rsh of a delta-doped epilayer of dia-
surement on sample De3 with an epilayer yields Rsh ¼ 20 kX/ mond grown on a defective substrate with an inhomogeneous
sq, which is conditioned by the low-ohmic edge region of the conductivity.
164503-7 Reznik, Korolyov, and Drozdov J. Appl. Phys. 121, 164503 (2017)

13
One of the primary claims of this work is that the R. Kantor and I. V. Shvets, J. Appl. Phys. 93, 4979–4985 (2003).
14
SNMM method enables spatially resolved characterization C. Gao, B. Hu, P. Zhang, M. Huang, W. Liu, and I. Takeuchi, Appl. Phys.
Lett. 84, 4647–4649 (2004).
of the sheet resistance of an epilayer when the supporting 15
D. Popovic, L. McCartney, C. Beasley, M. Lazebnik, M. Okonewski, S. C.
substrate has shunting and inhomogeneous conductivity. Hagness, and J. Booske, IEEE Trans. Microwave Theory Tech. 53,
With the method described in Ref. 20 (when only one 1713–1722 (2005).
16
Z. Wang, M. A. Kelly, Z.-X. Shen, L. Shao, W.-K. Chu, and H. Edwards,
unknown parameter of a test structure is to be determined),
Appl. Phys. Lett. 86, 153118 (2005).
such testing was possible only by independent scanning of a 17
A. Karbassi, C. A. Paulson, A. B. Kozyrev, M. Banerjee, Y. Wang, and D.
bare substrate and a substrate with the grown epilayer. W. van der Weide, Appl. Phys. Lett. 89, 153113 (2006).
18
However, the SNMM technique has potential for solving A. Imtiaz, S. M. Anlage, J. D. Barry, and J. Melngailis, Appl. Phys. Lett.
90, 143106 (2007).
more complicated inverse problems of testing structures with 19
W. Sun, Y. Yang, Z. Wu, T. Feng, Q. Zhuang, L.-H. Peng, S. Hu, and C.
a few parameters being measured simultaneously. The diag- K. Ong, J. Appl. Phys. 116, 044904 (2014).
20
nostics of multiparametric objects involves complex meas- A. N. Reznik and S. A. Korolyov, J. Appl. Phys. 119, 094504 (2016).
21
urements, for example, over some frequency ranges or using X.-D. Xiang and C. Gao, Mater. Charact. 48, 117–125 (2002).
22
A. N. Reznik and V. V. Talanov, Rev. Sci. Instrum. 79, 113708 (2008).
a number of probes differing in the aperture size. A possibil- 23
V. V. Talanov and A. R. Schwartz, IEEE Trans. Microwave Theory Tech.
ity of solving such problems with SNMM was demonstrated 57, 1224–1229 (2009).
24
in Ref. 47 through computer modeling. A. N. Reznik and E. V. Demidov, J. Appl. Phys. 113, 094501 (2013).
25
G. Gramse, M. Kasper, L. Fumagalli, G. Gomila, P. Hinterdorfer, and F.
Kienberger, Nanotechnology 25, 145703 (2014).
ACKNOWLEDGMENTS 26
A. Imtiaz, T. Baldwin, H. T. Nembach, T. M. Wallis, and P. Kabos, Appl.
The authors are grateful to colleagues from the laboratory Phys. Lett. 90, 243105 (2007).
27
CVD Diamond for Electronic Devices and Sensors, edited by R. S.
of Professor A. L. Vikharev (Institute of Applied Physics RAS, Sussmann (John Wiley & Sons, 2009).
N. Novgorod, Russia) for fabricating the diamond structures 28
H. Sumiya, N. Toda, and S. J. Satoh, Cryst. Growth 237–239, 1281–1285
and to Professor V. I. Shashkin (Institute for Physics of 29
(2002).
Microstructures RAS, N. Novgorod, Russia) for organization H. Sumiya and K. Tamasaku, Jpn. J. Appl. Phys., Part 1 51, 090102
(2012).
and technology provision of VDP-L measurements. We also 30
A. R. Lang, Diamond Relat. Mater. 2, 106–114 (1993).
thank the above-mentioned colleagues and Professor D. E. 31
F. Silva, J. Achard, O. Brinza, X. Bonnin, K. Hassouni, A. Anthonis, K.
Butler (Institute of Applied Physics RAS, N. Novgorod, De Corte, and J. Barjon, Diamond Relat. Mater. 18, 683–697 (2009).
32
P. M. Martineau, M. P. Gaukroger, K. B. Guy, S. C. Lawson, D. J.
Russia) for useful discussions. This work was supported by
Twitchen, I. Friel, J. O. Hansen, G. C. Summerton, T. P. G. Addison, and
RFBR under Grant No. 15-02-04081 and by the Program of R. Burns, J. Phys.: Condens. Matter 21, 364205 (2009).
33
the Department of Physical Sciences RAS. This research was J. Ristein, M. Riedel, M. Stammler, B. F. Mantel, and L. Ley, Diamond
based on the equipment of the Common Research Center Relat. Mater. 11, 359–364 (2002).
34
A. L. Vikharev, A. M. Gorbachev, M. A. Lobaev, A. B. Muchnikov, D. B.
“Physics and technology of micro- and nano-structures” at the Radishev, V. A. Isaev, V. V. Chernov, S. A. Bogdanov, and M. N.
Institute for Physics of Microstructures RAS. Drozdov, Phys. Status Solidi RRL 10, 324–327 (2016).
35
D. K. Schroder, Semiconductor Material and Device Characterization
1
B. T. Rosner and D. W. van der Weide, Rev. Sci. Instrum. 73, 2505–2525 (John Wiley & Sons, Inc., Hoboken, New Jersey, 2006).
36
(2002). J. Nahlik, I. Kasparkova, and P. Fitl, Measurement 44, 1968–1979
2
S. M. Anlage, V. V. Talanov, and A. R. Schwartz, “Principles of near-field (2011).
37
microwave microscopy,” in Scanning Probe Microscopy: Electrical and H. El-Hajj, A. Denisenko, A. Kaiser, R. S. Balmer, and E. Kohn, Diamond
Electromechanical Phenomena at the Nanoscale, edited by S. Kalinin and Relat. Mater. 17, 1259–1263 (2008).
38
A. Gruverman (Springer Science, New York, 2007). R. Edgington, S. Sato, Y. Ishiyama, R. Morris, R. B. Jackman, and H.
3
A. Imtiaz, T. M. Wallis, and P. Kabos, IEEE Microwave Mag. 15, 52–64 Kawarada, J. Appl. Phys. 111, 033710 (2012).
39
(2014). G. Chicot, T. N. T. Thi, A. Fiori, F. Jomard, E. Gheeraert, E. Bustarred,
4
C. Gao, T. Wei, F. Duever, Y. Lu, and X.-D. Xiang, Appl. Phys. Lett. 71, and J. Pernot, Appl. Phys. Lett. 101, 162101 (2012).
40
1872–1874 (1997). R. S. Balmer, I. Friel, S. Hepplestone, J. Isberg, M. J. Uren, M. L.
5
M. Tabib-Azar and Y. Wang, IEEE Trans. Microwave Theory Tech. 52, Markham, N. L. Palmer, J. Pilkington, P. Huggett, S. Hajdi, and R. Lang,
971–979 (2004). J. Appl. Phys. 113, 033702 (2013).
6 41
H. P. Huber, I. Humer, M. Hochleitner, M. Fenner, M. Moertelmaier, C. J. Scharpf, A. Denisenko, C. I. Pakes, S. Rubanov, A. Bergmaier, G.
Rankl, A. Imtiaz, T. M. Wallis, H. Tanbakuchi, P. Minterdorfer, P. Kabos, Dollinger, C. Pietzka, and E. Kohn, Phys. Status Solidi A 210, 2028–2034
J. Smoliner, J. J. Kopanski, and F. Kienberger, J. Appl. Phys. 111, 014301 (2013).
42
(2012). J. Pernot, P. N. Volpe, F. Omnes, P. Muret, V. Mortet, K. Haenen, and T.
7
A. Tselev, N. V. Lavrik, I. Vlassiouk, D. P. Briggs, M. Rutgers, R. Teraji, Phys. Rev. B 81, 205203 (2010).
43
Proksch, and S. V. Kalinin, Nanotechnology 23, 385706 (2012). E. A. Surovegina, E. V. Demidov, M. N. Drozdov, A. V. Murel, O. I.
8
J. Lee, C. J. Long, H. Yang, X.-D. Xiang, and I. Takeuchi, Appl. Phys. Khrykin, V. I. Shashkin, M. A. Lobaev, A. M. Gorbachev, A. L. Viharev,
Lett. 97, 183111 (2010). S. A. Bogdanov, V. A. Isaev, A. B. Muchnikov, V. V. Chernov, D. B.
9
D. E. Steinhauer, C. P. Vlahacos, S. K. Dutta, F. C. Wellstood, and S. M. Radishchev, and D. E. Batler, Semiconductors 50, 1569–1573 (2016).
44
Anlage, Appl. Phys. Lett. 71, 1736–1738 (1997). H. Kawarada, Surf. Sci. Rep. 26, 205–259 (1996).
10 45
C. Gao and X.-D. Xiang, Rev. Sci. Instrum. 69, 3846–3851 (1998). K. Hayashi, S. Yamanaka, H. Watanabe, T. Sekiguchi, H. Okushi, and K.
11
D. E. Steinhauer, C. P. Vlahacos, S. K. Dutta, B. J. Feenstra, F. C. Kajimura, J. Appl. Phys. 81, 744–753 (1997).
46
Wellstood, and S. M. Anlage, Appl. Phys. Lett. 72, 861–863 (1998). F. Liu, Y. Cui, M. Qu, and J. Di, AIP Adv. 5, 041307 (2015).
12 47
Y. Gao, A. Lauer, Q. Ren, and I. Wolf, IEEE Trans. Microwave Theory A. N. Reznik, I. A. Shereshevsky, and N. K. Vdovicheva, J. Appl. Phys.
Tech. 46, 1694–1703 (1998). 109, 094508 (2011).

You might also like