b - Anritsu

Report
Advances in mm-wave and THz Region
Materials Measurements
Jon Martens
Anritsu Company
EuMW Seminars 2013
Jeffrey Hesler
Virginia Diodes Inc.
1
Agenda
•
•
•
•
•
•
Background
Measurement systems
Quasi-Optical method
Lower mm-wave free-space method and waveguide methods
Mm-wave open coax method for liquids
Summary
EuMW Seminars 2013
2
Background
- Millimeter-wave materials measurements are increasingly needed
for electronic applications, better chemical and biological models,
for materials development, etc.
- Many lower frequency methods extend logically with reduced
sample size requirements but some things change
- Free-space and quasi-optical methods are more practical and can be
very useful
- S-parameter uncertainties change and this can affect extraction
- The materials parameters change which also affect uncertainty distributions
EuMW Seminars 2013
3
Anritsu measurement systems
The VectorStar VNA (20, 40, 50 and 70 GHz
versions) forms the basis for many mm-wave
measurements.
The base VNA can be combined with the 3743A
mm-wave modules for broadband coverage of 70
kHz-125 GHz…
… or used with VDI modules for much higher
frequency measurements.
EuMW Seminars 2013
4
Virginia Diodes – VDI Company Overview
• Company founded to respond to the needs
of the scientific community for THz sources
and receivers
– Founded in 1996, focused on diodes
– Reorganized in 2001, added components and
systems
– Originally focused primarily on astronomy,
spectroscopy and plasma diagnostics
– Field now expanding
• Imaging, radar, EPR/NMR, communications and
general THz test and measurement
• Developed a full range of broadband
electrically tunable solid state sources and
detectors
– Components from 50 GHz to 3 THz
– Ambient, no mechanical tuning
• Applying this technology to THz systems
– ESR/NMR Systems
– THz VNA Extenders
EuMW Seminars 2013
5
VDI - Core Technology
Use nonlinear diodes to extend the frequency range of
traditional microwave electronics
Microwave Technology
16.7 GHz
X3
VDI Technology
40
GHz
40
GHz
X8
20mW
1.5W
Schottky Diodes
CAD Design
Planar
– First-time design
– Broadband &
Tunerless
Advanced fabrication
technology
– High Efficiency
EuMW Seminars 2013
320
GHz
6
VDI VNA Extenders
• VDI Extenders available from WM-2540 (WR10, 75-110 GHz) thru WM-250 (WR-1.0, 7501050GHz)
• State-of-the-art Dynamic range
–
–
–
–
–
120 dB (typ.) at WM-2540 (WR-10, 70-110 GHz)
120 dB (typ.) at WM-1295 (WR-5.1, 40-220 GHz)
100 dB (typ.) at WM-380 (WR-1.5, 500-750 GHz)
70 dB (typ.) at WM-310 (WR-1.2, 600-900 GHz)
60 dB (typ.) at WM-250 (WR-1.0, 750-1050 GHz)
• Excellent amplitude and phase stability
EuMW Seminars 2013
7
THz Frequency Extension of a VNA
50 GHz
VNA
RF, LO & IF
Signal
Cables
THz
Extenders
Waveguide Test
Ports
EuMW Seminars 2013
8
WM-250 (WR-1.0) VNA Extender
•
•
•
•
•
•
•
•
Dynamic Range: 60 dB typical at 10Hz BW
Dynamic Range: 40 dB minimum at 10Hz BW
Magnitude Stability: ±1 dB
Phase Stability: ±15°
Test Port Power: -35 dBm
Test Port Input Limit (dBm, saturation/damage): 20/13
Directivity: 30 dB
Typical Dimensions: 8 x 5 x 3 inches
EuMW Seminars 2013
9
THz Waveguide Calibration
• mm-Wave waveguide calibration
methods
– Short-Open-Load-Thru (SOLT, TOSM)
• Open typically uses ¼-wave delayed
short
– Thru-Reflect-Line (TRL, LRL)
• Line is typically a ¼-wave thru shim
– Many others possibilities as well…
• Sub-mm wave introduces a new
set of challenges
– Thru-Reflect-Line (TRL, LRL)
Machined Quarter-wave
Delay Short
• ¼-wave shims difficult to fabricate and
fragile
• Common to instead use two lines with
¼-wave difference in length
– However, this means more connections and
interfaces less accuracy
– Short-Open-Load-Thru (SOLT, TOSM)
• Challenging to achieve a high return
loss load
EuMW Seminars 2013
10
WM-380 (WR-1.5) Precision Loads
EuMW Seminars 2013
11
WM-250 (WR-1) Calibrated Measurements
• Calibrated waveguide measurements were performed using
the Extender
• Short-Open-Load-Thru calibration method
– ¼-wave delay was used as the Open standard
• 1 kHz IF Bandwidth for calibrated measurements
Precision Load
EuMW Seminars 2013
Quarter-wave Delay
12
WR-1 Waveguide
Milled Delay
Quasi-Optical Measurements
Anritsu
VectorStar
VDI WM-250 (WR-1)
Extender
Test Port with
Feedhorn
Reflecting
Plate
Off-Axis Parabolic
Focusing Mirrors
EuMW Seminars 2013
13
Two-Port Quasi-Optical Measurements
2-Port Measurement in Collimated Beam
Feedhorn
Feedhorn
Lens
DUT
Lens
2-Port Measurement in Focused Beam
EuMW Seminars 2013
• Requires large
sample
• Sample in
collimated
beam
14
• Can use small
sample
• Good for
devices (grid
arrays, focused
optics, …)
• Difficult
Alignment
One-Port Quasi-Optical Measurements
1-Port Measurement in Collimated Beam
Reflecting
Plate
Feedhorn
Lens
DUT
• Can use small
sample
• Good for
devices (grid
arrays, focused
optics, …)
1-Port Measurement in Focused Beam
EuMW Seminars 2013
• Requires large
sample
• Sample in
collimated
beam
15
Quasi-Optical Measurement Calibration
Measurement of Sample in Focused Beam
• Slide lens and sample as a unit
– Translation occurs in collimated section
– Improved accuracy, minimal effect on focusing
• e.g. see Arsenovic 2013 MTT
– Measure multiple delay lengths
EuMW Seminars 2013
16
One-Port Quasi-Optical Measurement Setup
Reflecting
Plate
Test Port with
Diagonal
Feedhorn
Device
Under Test
taped to
plate
Collimated
Beam
EuMW Seminars 2013
17
Quasi-Optical Measurement Setup
Focusing
Mirror and
Sample move
as a unit
EuMW Seminars 2013
18
Over-Determined Least Squares Calibration
• For calibration, use an un-weighted least squares method
– e.g. see Wong 2004 ARFTG
• Calibration standards
– Series of delayed reflections with known delay distances
• Metal plate at focal point of QO system
• Metal plate and focusing mirror are both mounted on the same moving stage
– Matched load
• Used absorber at 45 degree angle as rough termination
– Measurement bandwidth 300 Hz
EuMW Seminars 2013
19
Over-determined Least Squares Calibration
EuMW Seminars 2013
20
Evaluation of Calibration Quality
• Look at Sparameters of the
calibration
standards
– Over-determined
calibration  a
measure of the
calibration quality
• Gives an indication
of the measurement
quality that can be
achieved
EuMW Seminars 2013
21
S-Parameters of Quasi-Optical System
• A measurement of
the S-parameters of
the Quasi-Optical
system
• Measured using a
two-tiered extraction
method
• Uses a waveguide
calibration followed
by a separate QO
calibration
Water Line
–
See Arsenovic 2013
MTT
• Vector Star has
onboard two-tiered
calibration available
–
EuMW Seminars 2013
22
Network Extraction
Measurements of Dielectric Samples
• Samples were taped to
the metal plate at the
focal point
• Repeated measurements
were made of each
sample
0.014” Aluminum NItride
– Sample removed from
plate and re-mounted for
each measurement
1 mm Quartz (Fused Silica)
• Calibration using the QO
method described earlier
– Not a two-tiered
calibration
EuMW Seminars 2013
23
0.014” Aluminum Nitride
EuMW Seminars 2013
24
0.014” Aluminum Nitride
EuMW Seminars 2013
25
0.014” Aluminum Nitride
EuMW Seminars 2013
26
0.014” Aluminum Nitride
EuMW Seminars 2013
27
1 mm Quartz (Fused Silica)
EuMW Seminars 2013
28
1 mm Quartz (Fused Silica)
EuMW Seminars 2013
29
1 mm Quartz (Fused Silica)
EuMW Seminars 2013
30
1 mm Quartz (Fused Silica)
EuMW Seminars 2013
31
Lower mm-wave free-space
• TRL calibrations are another option; particularly in the 100 GHz range
• The position for the Line is not critical as long as under a halfwavelength (1.5mm at 100 GHz). A stop on the translation stage allows
a relatively precise return to the main reference plane.
• A shorting plate at the reference plane can be the reflect. The
sensitivities to non-planarities here are surprisingly low.
• Multipath reflections are still a larger concern.
Main reference plane
Translation stage to
position L
To VNA port 1
EuMW Seminars 2013
Absorber
32
To VNA port 2
mm-wave free-space
Free-space TRL, effect of reflect asymmetry
-10
-20
|S11| (dB)
• Fairly large reflect
asymmetries have relatively
small impact. Of course, it
could be tilted enough to
make it into a load
standard….
20% mag and phase asymmetry
symmetric
-30
-40
-50
-60
70
75
80
85
90
95
100
105
110
Frequency (GHz)
To
port 1
EuMW Seminars 2013
Multipath can
be an issue
without
absorbers or
another
strategy.
To
port 2
33
mm-wave free-space: cellulose example
• In the first measurement, multipath control and focusing were only
partially adequate. Uncertainties increased…
• Gating does help again (gate width here ~1 cm) .
free-space celluose measurement
0
-0.2
2
real
real (gating)
imag
imag (gating)
-0.4
1
-0.6
55
EuMW Seminars 2013
65
75
85
Frequency (GHz)
34
95
imaginary permittivity
real permittivity
3
mm-wave free-space: thin conductor example
• A conductivity estimate can be extracted from the complex permittivity
assuming a simple transport model. For disordered materials, this can cause
deviations from DC expectations. Proper reference plane positioning does
help (~20mm accuracy in the ‘better’ case).
Contaminated metal film conductivity estimate
|S11| of order unity
k
S  S 1
2
11
2
21

S 1
2 S 11
2
11
2 S 11
Kernel from
Nicolson-Ross-Weir
EuMW Seminars 2013
Est. conductivity (MS/m)
2.5
2
1.5
1
ref plane error
DC
0.5
better ref planes
0
55
60
65
35
70
75
80
Frequency (GHz)
85
90
95
100
Mm-wave waveguide/fixtured methods
• Unlike free-space, waveguide setups offer fewer
modal/multipath issues but sample size is more constrained.
• In transmission line-like methods, the S-parameters are usually
related to the materials parameters (G=G(e,m), g=g(e,m)) via:
S 11  e
 2 g 0 L1
 G 1  e


2  2 gd 
1

G
e


2 gd
L1
S 21
d
 gd
2




e
1

G
 g 0  L1  L 2 
 e

2  2 gd 
1

G
e


L2
er, mr
• Depending on how well-known the reference plane positions
and sample thickness are, simplifications are possible.
EuMW Seminars 2013
36
Background: S-parameter uncertainties
At mm-wave frequencies, noise floors degrade, repeatability
becomes more challenging and calibration material knowledge may
decrease. Net uncertainties tend to degrade, particularly for highly
lossy materials.
Transmission magnitude uncertainty
Transmission magnitude uncertainty
10
1
1 GHz
0.1
dB
dB
1
80 GHz
750 GHz
0.1
900 GHz
110 GHz
145 GHz
0.01
-80 -60 -40 -20
0
Transmission coefficient (dB)
-80 -60 -40 -20
0
Transmission coefficient (dB)
EuMW Seminars 2013
1100 GHz
0.01
37
Waveguide/fixtured methods: uncertainty
• Uncertainty propagation:
– Sample length dependency tends to scale roughly with l/L ; decreasing
wavelength helps for fixed size
Relative permittivity uncertianty
permittivity uncertainty estimate
Transmission dominant
measurement of low-loss material;
permittivity decreasing from 5 to 2;
fixed length uncertainty of 5mm
0.5
0.4
10 GHz
0.3
100 GHz
0.2
0.1
0
0
0.5
1
Sample length (mm)
1.5
2
While absolute S-parameter uncertainties increase with frequency, permittivities tend to
decrease…these can sometimes balance out
EuMW Seminars 2013
38
Waveguide/fixtured example: slushy sea water
in W-band
• Precise temperature control can be critical for some
measurements
• The use of thin walled ‘thermal break’ waveguide sections and
radiative blocks can help get << 1C control
• Use of membrane-cal fixtures helps get reference planes correct
Rest of thermal wrap not
shown here…
Slush sample in the middle
waveguide section
Waveguide calibration at the
break interfaces.
EuMW Seminars 2013
39
Waveguide/fixtured methods: example
• With decent temperature control and proper calibration
planes, one can get close to expected results…
(35%) salt water permittivity vs. temp
7
6
(35%) salt water permittivity vs. temp
5
4
0C model
10C model
0C meas
10C meas
-4
Imagninary permittivity
Real permittivity
8
3
75
80
85
90
95
Frequency (GHz)
100
105
110
-6
-8
0C model
10C model
0C meas
10C meas
-10
-12
-14
75
80
Basic S-parameter component of the uncertainty here is <0.05.
EuMW Seminars 2013
40
85
90
95
Frequency (GHz)
100
105
110
Waveguide/fixtured methods: incomplete fill
• When sample size is not ideal, aperture effects can be deembedded. Often these effects are model-able with good accuracy.
Unfilled WG
aperture
Iris transmission
-20
sample
|S21| (dB)
-30
-40
-50
measurement
-60
model
-70
65
EuMW Seminars 2013
41
75
85
Frequency (GHz)
95
105
Waveguide conducting film measurement
example
• A small gap can have a significant impact, particularly for highly
lossy/conductive samples.
• S11 phase is the dominant measurement.
Expected value range based
on other measurements…
Thin conducting sample measurement
200000
150000
Effect of S11 phase on thin conductor evaluation
(W-band)
100000
w/ de-embedding
50000
w/o de-embedding
0
70
80
EuMW Seminars 2013
90
100
Frequency (GHz)
110
120
Extracted conductivity (S/m)
Extracted conductivity (S/m)
250000
30000
20000
10000
0
-177
42
-175
-173
S11 phase (deg)
-171
Mm-wave open-ended coax
• Also used for years at lower frequencies, this method has
become of increasing interest for studies of liquids at higher
frequencies
– Multiple time constants for different mechanisms
– Better model fitting
1mm connector;
110 GHz (125 GHz)
1.85mm connector; 70 GHz
EuMW Seminars 2013
43
Open-ended coax
• Coax diameter (b) tends to determine upper frequency limit
while probe size (c) tends to set the lower frequency limit
(together with side structure)
b
c
• A sealed glass bead at the interface
keeps porosity and contamination
possibilities low.
• A variety of plating and/or base
metal choices for the probe body
can be chosen for more challenging
environments. Default is goldplated aluminum.
b< 1mm for the 125 GHz probe, >1mm for the 70 GHz probe
EuMW Seminars 2013
44
Open-ended coax
• In the simplest analysis, one can integrate over the aperture to
get an expression for the admittance*.
Y  F  , a , b , e 
b
EuMW Seminars 2013
a
• We need at least 3 admittance
standards at the aperture to
calibrate. Three media of known e
will do.
• Air, water and a shorting medium
are commonly used.
• High order modes (particularly at
mm-wave) complicate things but,
through modeling, can be
corrected on a probe-type basis.
*J. Hunger, Max Planck Institute for Polymer Research
45
Open-ended coax
• Measurement of water one day after initial calibration. Since
this was a calibration standard, this is more of a calibration
stability test. Values returned to <<1%.
EuMW Seminars 2013
46
Open-ended coax
• Broadband isopropanol example (20 mL sample) in four
measurement trials
Real permittivity of isopropanol
11
permittivity
9
7
a
b
c
d
ref
Imaginary permittivity of isopropanol
5
0
3
-1
1
21
41
61
81
Frequency (GHz)
101
121
Imag permittivity
1
-2
-3
-4
a
b
c
d
ref
-5
-6
-7
1
EuMW Seminars 2013
47
21
41
61
81
Frequency (GHz)
101
121
Open-ended coax: repeatability
• Sample-to-sample repeatability can be <<1%.
real isopropanol permittivity: sample-tosample
imaginary isopropanol permittivity:
sample-to-sample
25
permittivity
20
permittivity
0
trial 1
15
trial 2
10
trial 3
-2
-4
trial 1
-6
trial 2
-8
5
trial 3
-10
0
0.1
20.1
40.1
Frequency (GHz)
0.1
60.1
20.1
40.1
Frequency (GHz)
60.1
• Cal-to-cal repeatability can also be good but may be limited by DI water purity
imaginary isopropanol permittivity: cal-tocal
real isopropanol permittivity: cal-to-cal
0
permittivity
20
permittivity
25
trial 1
15
trial 2
10
trial 3
-2
-4
trial 1
-6
trial 2
-8
5
trial 3
-10
0
0.1
20.1
40.1
Frequency (GHz)
EuMW Seminars 2013
0.1
60.1
48
20.1
40.1
Frequency (GHz)
60.1
Open-ended coax: repeatability and probe
position
• In a small volume container, results can still be reasonably consistent
staying at least 5 coax radii from the walls (~1mm here)
Real permittivity of isopropanol vs. probe position
Open top
9
a
b
c
7
permittivity
b
c
d
5
e
a
f
g
e
3
d
f
g
Position of aperture
relative to container
1
1
EuMW Seminars 2013
10
Frequency (GHz)
100
49
Open-ended coax: short repeatability
• A fine-pore metallic foam, very soft foil or silver paint can make a
good short standard. A harder foil can reduce repeatability,
particularly at low frequencies where capacitive coupling is weak.
effect of short contact
Re(Water permittivity)
160
120
80
hard foil
soft
40
0
0.1
EuMW Seminars 2013
1
10
Frequency (GHz)
50
100
mm-wave range
repeatability is, oddly,
easier to achieve
Conclusions
• Full two-port VNA Extenders with excellent dynamic range are
available up to 1.1 THz. Broadband measurements are possible
from near DC to 125 GHz in one connection.
• Quasi-optical one-port measurements were successfully
demonstrated in the WM-250 (WR-1) frequency band (750-1100
GHz)
– A simple method was used to mount the samples into the system
• The sample holding and alignment could be improved for better measurements
– Dielectric constant and loss tangent of the samples could be determined
• Other techniques also apply including mm-wave versions of
fixtured methods and open-ended coax probes
– Small sample sizes are possible with good repeatability
– Thin conducting and liquid samples can be handled particularly well
EuMW Seminars 2013
51
Additional material
EuMW Seminars 2013
52
Technology Example: Frequency Multiplier
WR-2.8 Tripler
Output
265-400 GHz
Input
88-133 GHz
EuMW Seminars 2013
53
• Tunerless
• Ambient operation
• Rugged and
repeatable
VDI THz Sources : 220-330 GHz
Synthesizer Extender
• WR-3.4 (220-330 GHz) Frequency Extender for
Synthesizers
–
Tunerless, instantaneous sweeping over > 40%
bandwidth
• AM modulation and Power Control capability
–
–
X3
X3
Input
< 20 GHz
EuMW Seminars 2013
Voltage controlled
Can also be controlled by drive synthesizer
X3
Output
220-330 GHz
54
Measurement of Waveguide Attenuator
EuMW Seminars 2013
55
Evaluation of Calibration Quality
Delayed
Shorts
Load
EuMW Seminars 2013
56
Measurement of 1” Waveguide Straight
Ideal loss X 1.5 to
account for surface
roughness
EuMW Seminars 2013
57
Evaluation of Calibration Quality - Shorts
EuMW Seminars 2013
58
WR-1.0 – Amplitude & Phase Stability
•
Look at amplitude & phase stability of system over one hour
–
–
–
•
Stability was measured in general laboratory space
–
–
EuMW Seminars 2013
Stability is important to maintain the calibration during the measurements
Measured for full 2-port WR-1.0 extender
1-port stability typically 5-10 times better
Poorly controlled thermal environment
Significantly improved performance can be achieved in a controlled thermal
environment
59
Atmospheric Transmission
1E+03
Loss (dB/m)
1E+02
1E+01
1E+00
1E-01
1E-02
1E-03
750
EuMW Seminars 2013
850
950
1050
Frequency (GHz)
60
0.014” Aluminum Nitride
EuMW Seminars 2013
61

similar documents