Development of NSTAP: Nanoscale Thermal

Anemometry Probe

Gary J. Kunkel∗, Craig B. Arnold†, and Alexander J. Smits‡

Princeton University, Princeton, NJ, 08544, US

A nanoscale thermal anemometry probe (NSTAP) has been developed to measure in-stantaneous fluid velocity at ultra-small scales. The sensing length of the current probe(∼60 µm) is and order of magnitude smaller than presently available commercial hot-wireanemometer probes (TSI Inc, Dantec Inc). The sensing element is a freestanding nanowire100 nm × 1µm × 60 µm suspended between two current-carrying contacts. The probeis constructed using standard semiconductor and microelectromechanical systems manu-facturing methods. The increased surface area to volume ratio of the metallic nanowirein comparison to conventional probes yields a device that not only has a higher spatialresolution but is also more sensitive and rapid in its response to changing flows.

Nomenclature

E Voltage across wireEO Output voltage of circuitI Wire currentR1 Voltage divider fixed resistor valueRw Resistance of wireRoff Resister used to set output voltage offset

I. Introduction

A nanoscale thermal anemometry probe (NSTAP) has been designed and developed to measure instanta-neous fluctuating velocities at ultra-small scales. The operating principle of the NSTAP is the same as thatused in current thermal anemometry.1 Using a simple electronic circuit (e.g. a voltage divider) an electriccurrent (∼mA) is passed through a small wire heating it above the ambient fluid temperature. As the fluidflows over the wire, the convective heat transfer from the wire to the fluid cools the wire, which changes itsresistance. This resistance change is monitored with an electronic circuit which provides an electric signalthat is a function of the flow velocity. Once the anemometer is calibrated, it can be used to measure thevelocity of the surrounding fluid.

Ligrani and Bradshaw2,3 have used conventional hot-wire construction techniques to make small wires,while others4–7 have incorporated other microelectromechanical manufacturing techniques. Such approacheshave been successfully used to measure turbulent flows; however, none are capable of studying ultra-small-scale turbulence. With a sensing element two orders of magnitude smaller than current commercial tech-nology, the spatial and temporal resolution of the NSTAP greatly exceeds existing commercial hot-wireanemometers. These new probes have measuring lengths less than ∼ 20 µm and widths less than ∼ 100 nm.For comparison, conventional hot-wire probes typically have a physical measuring length of ∼ 1 mm and adiameter of 5 µm. Figure 1 shows a size comparison of a conventional hot-wire probe and NSTAP with a0.1 µm × 1 µm × 60 µm wire before focused ion beam milling.

∗Research Staff, Department of Mechanical and Aerospace Engineering, Member AIAA†Professor, Department of Mechanical and Aerospace Engineering‡Professor, Department of Mechanical and Aerospace Engineering, Fellow AIAA

1 of 8

American Institute of Aeronautics and Astronautics

36th AIAA Fluid Dynamics Conference and Exhibit5 - 8 June 2006, San Francisco, California

AIAA 2006-3718

Copyright © 2006 by Gary J. Kunkel. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

Dow

nloa

ded

by P

RIN

CE

TO

N U

NIV

ER

SIT

Y o

n Ja

nuar

y 20

, 201

8 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

006-

3718

a) Conventional 5 m hot wire b) NSTAP

c) NSTAP with .1 m x 1 m x 60 m wire

50x

500x

d) Scanning electron micrograph

Figure 1. Visual comparison of NSTAP and conventional hot-wire probe.

2 of 8

American Institute of Aeronautics and Astronautics

Dow

nloa

ded

by P

RIN

CE

TO

N U

NIV

ER

SIT

Y o

n Ja

nuar

y 20

, 201

8 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

006-

3718

The probe has been developed to study the statistical and structural nature of ultra-small-scale turbu-lence, specifically at high Reynolds numbers. The Reynolds number, representing the ratio of the largestto smallest scales in the flow, is the most significant non-dimensional parameter of turbulent flow. Manypractical flows, such as the flow over a large vehicle, or the flow in the earth’s atmosphere, are characterizedby very high Reynolds numbers. The effect of the Reynolds number on these flows must be understood ifwe are to understand the physics of the flow as a whole.

As an example, ongoing work at Princeton has shown that it is possible to reach very high Reynoldsnumbers in a laboratory-sized facility by using high-pressure air as the working fluid. For instance, thePrinceton/ONR Superpipe operates at pressures up to 3,500 psi, increasing the density by factors up to 240over its ambient value. This approach yields pipe flow Reynolds numbers up to 38×106. However, while thesefacilities allow for high Reynolds number testing in a laboratory setting, the major disadvantage of any highReynolds number laboratory facility is the inability of conventional instrumentation to resolve the smallestturbulent energy-containing scales. With a fixed outer-scale of the flow (boundary layer thickness or piperadius), an increase in Reynolds numbers necessarily means a decrease in the physical size of the smallestscales of turbulence. For example, at the highest Reynolds numbers obtained in the Superpipe, the ratio ofthe size of the largest-scale motions to the smallest-scale motions is of the order of 106. This means thatthe smallest eddies are of micron size. In the study of turbulence structure, it is essential to determine thecharacteristics of such small scale motions accurately, which presents a considerable experimental challengerequiring an extremely small and fast probe. The NSTAP will enable the accurate measurement of suchsmall-scale turbulent phenomena by using innovative micromachining technology and will allow small-scalefacilities to open up an entirely new frontier in the study of high-Reynolds-number turbulent flow. Using theNSTAP, spatially and temporally resolved turbulence data could be obtained at Reynolds numbers matchingthe flow over a commercial aircraft and in large pipe lines. These results will lead to more accurate predictionsof the behavior of such technologically important flows.

Note that while the immediate use for the NSTAP is the study of small-scale turbulence, the constructionof a freestanding nanowire with aerodynamic supports has many other uses. For instance, nanowires havebeen used as nanosensors for biological and chemical species8 and as nanoscale interconnects,9 and otherresearchers suggest that the many uses of nanowires are limited only by ones imagination.10 Furthermore,with the growing field of micro-fluidics devices,11 sensors such as the NSTAP could provide a complementaryanalysis to existing instrumentation such as micro-PIV.

II. Construction

The main design concept for the NSTAP consists of using standard semiconductor and microelectro-mechanical systems manufacturing techniques to produce a freestanding nanowire between two electricallyconducting support posts. The fabrication procedure is divided into four main processes:

1. Oxidation and photolithography

2. Metal deposition and backside chemical etching

3. Focused Ion Beam (FIB) milling

4. Precision Laser Micromachining (PLM) and chemical etching release of nanowire

The steps in the manufacturing process are shown in figure 2. We start by growing a ∼ 500 nm thicklayer of SiO2 using wet thermal oxidation. The oxide provides an insulating layer between the metal and thesilicon substrate. In addition, the oxide layer performs the important duty of supporting the metal nanowireduring processing prior to device activation. Standard optical lithographic techniques are used to patternthe probe on top of the oxide (front-side).

An electron beam evaporator is used to deposit a 10 nm layer of titanium on the front side of the wafer(to promote adhesion) followed by a 100 nm thick layer of platinum (note, this thickness of the platinum isthe thickness of the nanowire perpendicular to the flow). After deposition, the unwanted platinum is liftedoff using a solution of PRS-1000 at 100 ◦C. The deposited film is thermally annealed to remove any intrinsicstresses generated in the film through the deposition process. The patterned probe looking from the top sideof the wafer is shown in figure 3. The probe consists of two 5 × 1 mm contact pads narrowing at the frontof the probe (to reduce the aerodynamic interference) to a 0.1 µm × 1 µm × 60 µm wire.

3 of 8

American Institute of Aeronautics and Astronautics

Dow

nloa

ded

by P

RIN

CE

TO

N U

NIV

ER

SIT

Y o

n Ja

nuar

y 20

, 201

8 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

006-

3718

(a)

(b)

(c)

(d)

(e)

(f)

Figure 2. Schematic of fabrication procedure: (a) bare silicon wafer, (b) growth of thin (500 nm) oxide layer,(c) deposition of 100 nm Pt layer, (d) photolithography and chemical etching to create large scale structure ontop and bottom of wafer, (e) large wire has been milled in the FIB to generate the nanowire at tip of probe andoxide is etched to release the nanowire, (f) entire probe is laser micromachined to produce the aerodynamicstructure. The dashed line represents the location at which the cross-section has been taken.

A focused ion beam can then be used to mill the patterned wire down to the desired sensing length. Theamount of milling performed determines the length and depth of the nanowire (the sensing element of theprobe). For the current generation of probes, the nanowire has been milled to 0.1 µm × 0.1 µm × 20 µm.We expect to construct smaller wires in the future. The stubs between the wire and the probe pads are usedto thermally separate the hot wire from the relatively large probe contact pads.12 The next step, which weare currently working on, is to use precision laser micromachining to remove the probes from the wafer andto produce an aerodynamic sensor. The SiO2 holding the nanowire is then removed from between the probepads and below the wire using a front-side dry reactive-ion etch of CH4 and of H2. Then the Si between theprobe pads is removed using SF6 and O2. Upon completion, we have a freestanding wire (0.1 µm × 1.0 µm× 60 µm).

III. Operation

Once the probe is constructed it can be used in a standard constant-temperature anemometer or in aconstant-current anemometer bridge circuits.13 For the results presented here, a 0.1 µm × 1 µm × 60 µmfreestanding wire (figure 1 d) is operated with the simple bridge circuit shown in figure 4. Note, as shown in

4 of 8

American Institute of Aeronautics and Astronautics

Dow

nloa

ded

by P

RIN

CE

TO

N U

NIV

ER

SIT

Y o

n Ja

nuar

y 20

, 201

8 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

006-

3718

Figure 3. Picture of the NSTAP on front side of wafer at step (d) in figure 2. The sensing element (locatedat the right edge of the probe) is not visible in this view.

figure 1 d), that the test wire is not at the immediate end of the probe. That is, ∼100 µm of the remaining Siprotrudes out alongside of the freestanding wire. While this wire is not expected to give accurate quantitativeflow measurements it is used to test the qualitative response of the wire to fluctuating velocities.

1

2

3

4-

+R1R1

RoffRw

E

Figure 4. Electronic testing circuit for the NSTAP. Here the Op-amp is and instrumentation amplifier.

The current-voltage characteristic curve from the NSTAP and a conventional hot wire operated with thesame circuit are shown in figure 5. The conventional hot-wire has a sensor diameter of 5 µm and a sensorlength of 1 mm. Qualitatively, the trends are similar, with the NSTAP having only a slightly smaller changein resistance with changing voltage across the wire.

The static response of the wire to a varying flow up to ∼ 25 m/s is shown in figure 6. For all datashown here the wires are operated at an overheat ratio of 1.18. The NSTAP is seen to respond similarly to aconventional hot-wire. Figure 7 shows the power spectral densities of the NSTAP as well as a conventionalhot wire operated with the test circuit in figure 4 and the hot wire operated with a commercial constant-temperature anemometer. Qualitatively, the NSTAP is seen to have a broader dynamic response than thelarger conventional hot wire when operated with the same circuit.

IV. Conclusions

Standard semiconductor processing techniques have been used to construct a freestanding 0.1 µm × 1 µm× 60 µm platinum wire. The wire has been operated with a constant current bridge circuit at an overheatratio of 1.18 and was found to have a similar response to the flow as a conventional wire operated with thesame circuit. The wire was also placed in a turbulent flow and was found to have a broader response than it’sconventional counterpart. From these tests the current construction procedures seem promising and workhas begun on the next generation of probes. The next step in the probe development is to use precisionlaser micromachining to remove the bulk silicon from underneath and along side of the probe resulting in anaerodynamic structure.

5 of 8

American Institute of Aeronautics and Astronautics

Dow

nloa

ded

by P

RIN

CE

TO

N U

NIV

ER

SIT

Y o

n Ja

nuar

y 20

, 201

8 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

006-

3718

0 0.2 0.4 0.6 0.8 1

E/Emax

0

0.5

1

I/Imax

hot wire

NSTAP

=I/Imax E/E

max

Figure 5. Current-voltage characteristics of the NSTAP and conventional hot-wire in circuit shown in figure4. Both probes were operated at an overheat ratio of 1.18 and have been normalized with their maximumvalues. The solid line is the NSTAP and the dashed line is a conventional hot-wire.

6 of 8

American Institute of Aeronautics and Astronautics

Dow

nloa

ded

by P

RIN

CE

TO

N U

NIV

ER

SIT

Y o

n Ja

nuar

y 20

, 201

8 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

006-

3718

0 5 10 15 20 25

U (m/s)

0

0.5

1

1.5

2

2.5

3

3.5

4

EO (V)

NSTAP

hot wire (CC)

Figure 6. Example calibration of the NSTAP (circles) and the conventional hot wire (squares). Both wereoperated with the circuit in figure 4 at an overheat ratio of 1.18.

References

1Bruun, H. H., Hot-wire Anemometry, Oxford University Press, Oxford, UK, 1995.2Ligrani, P. M. and Bradshaw, P., “Spatial resolution and measurement of turbulence in the viscous sublayer using

subminiature hot-wire probes,” Experiments in Fluids, Vol. 5, 1987, pp. 407–417.3Ligrani, P. M. and Bradshaw, P., “Subminiature hot-wire sensors: development and use,” J. Phys. E: Sci. Instrum.,

Vol. 20, 1987, pp. 323–332.4Jiang, F., Tai, Y.-C., Ho, C.-M., Karan, R., and Garstenauer, M., “Theoretical and experimental studies of micromachined

hot-wire anemometers,” International Electron Devices Meeting, San Francisco, CA, IEEE, New York, NY, 1994, pp. 139–142.5Ho, C. M., Li, W., Garstenauer, M., Karan, K., Leu, T.-S., and Tai, Y.-C., “Micromachined Hot-Point Anemometer.

Part II: Testing and Calibration,” Bull. American Physical Soc., Vol. 38, 1993, pp. 2234.6Tai, Y.-C., Jiang, F., Liu, C., Wu, R., and Ho, C. M., “Micromachined Hot-Point Anemometer. Part I: Design and

Fabrication,” Bull. American Physical Soc., Vol. 38, 1993, pp. 2234.7Naguib, A., Benson, D., Nagib, H., Huang, C., and Najafi, K., “Assessment of new MEMS-based hot wires,” Proc. 3rd

ASME/JSME Joint Fluids Engineering Conference, ASME, New York, NY, 1999, pp. 1–8, FEDSM99-7355.8Cui, Y., Wei, Q., Park, H., and Leiber, C. M., “Nonowire nanosensors for highly sensitive and selective detection of

biological and chemical species,” Science, Vol. 293, 2001, pp. 1289–1292.9Wu, B., Heidelberg, A., and Borland, J. J., “Mechanical properties of ultrahigh-strength gold nanowires,” Nature Mate-

rials, Vol. 4, 2005, pp. 525–529.10Appell, D., “Nanotechnology: Wired for success,” Nature, Vol. 419, 2002, pp. 553–555.11Gad-El-Hak, M., The MEMS Handbook , CRC Press, 2001.12Li, J. D., McKeon, B. J., Jiang, W., Morrison, J. F., and Smits, A. J., “The response of hot wires in high Reynolds-number

turbulent pipe flow,” Meas. Sci. Technol., Vol. 15, 2004, pp. 789–798.13Huang, J. B., Jiang, F. K., Tai, Y. C., and Ho, C. M., “A micro-electro-mechanical-system-based thermal shear-stress

sensor with self-frequency compensation,” Meas. Sci. Technol., Vol. 10, 1999, pp. 687–696.

7 of 8

American Institute of Aeronautics and Astronautics

Dow

nloa

ded

by P

RIN

CE

TO

N U

NIV

ER

SIT

Y o

n Ja

nuar

y 20

, 201

8 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

006-

3718

101

102

103

104

f (Hz)

10-11

10-10

10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

PS

D (

arb

itra

ry u

nit

s)

NSTAP (CC)

hot wire (CT)

hot wire (CC)

Figure 7. Example power spectral density of the NSTAP, a conventional hot wire operated with the testcircuit in figure 4, and the hot wire operated with a commercial constant-temperature anemometer.

8 of 8

American Institute of Aeronautics and Astronautics

Dow

nloa

ded

by P

RIN

CE

TO

N U

NIV

ER

SIT

Y o

n Ja

nuar

y 20

, 201

8 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

006-

3718