sniffers: fluid-dynamic sampling for olfactory trace ... · sniffers is the principal aim of this...

ctoryerte-Somecemeal,ever,o detectround,iffing.lec-them

tection,g ap-everalmics.broad

Gary S. Settlese-mail: [email protected]

Department of Mechanical and NuclearEngineering, Penn State University,

University Park, PA 16802

Sniffers: Fluid-Dynamic Samplingfor Olfactory Trace Detection inNature and HomelandSecurity—The 2004 FreemanScholar LectureVertebrates aim their noses at regions of interest and sniff in order to acquire olfatrace signals that carry information on food, reproduction, kinship, danger, etc. Invbrates likewise position antennae in the surrounding fluid to acquire such signals.of the fluid dynamics of these natural sensing processes has been examined piebut the overall topic of sniffing is not well investigated or understood. It is, howimportant for several human purposes, especially sampling schemes for sensors tchemical and biological traces in the environment. After establishing some backga general appraisal is given of nature’s accomplishments in the fluid dynamics of snOpportunities are found for innovation through biomimicry. Since few artificial (“etronic”) noses can currently sniff in the natural sense, ways are considered to helpsniff effectively. Security issues such as explosive trace detection, landmine dechemical and biological sniffing, and people sampling are examined. Other sniffinplications including medical diagnosis and leak detection are also considered. Sresearch opportunities are identified in order to advance this topic of biofluid dynaThough written from a fluid dynamics perspective, this review is intended for aaudience.fDOI: 10.1115/1.1891146g

marindlebuturinebratheeorald

tist

d t

orenthtiotly

y ithic

eed

lik.S

Lab,

thatther

an ato-

fluidspic.trike

uid-e-lves

argetAlso,

mples

tec-een

r air-to be

that

e airnts

ion,

1 IntroductionSniffing for chemical traces is as old as nature itself. It do

nates the lives of most animal species, relegating touch, heand sometimes even vision to lower status. Animals depenolfaction for food, reproduction, kin recognition, and danger af1g. The first three of these require little sniffing in humans,recognizing danger is still very important to us in the 21st cen

The study of olfaction is a well-established scientific disciplbut hardly a stagnant one. Zwaardemaker’s book on verteolfaction f2g already listed over 200 references in 1895, but atwriting, researchers R. Axel and L. B. Buck have just bawarded the Nobel Prize in Medicine for their research on odreceptor genetics. Some of the many recent surveys of the fiegiven in Refs. 1 and 3–6. These authors discuss olfaction fromperspectives of neuroscience, psychology, zoology, chemanatomy, and communication. Few, however, have includeperspective of fluid dynamics.

But fluid dynamics is central to olfaction, and it is therefsurprising that sniffing flows have not attracted much attentiodate. Biofluid dynamics has been mainly concerned withlarger themes of blood flow, respiration, and external locomoflows f7,8g, and has left the fluid dynamics of sniffing mosunexplored.

Hence the time is right to investigate this topic, especiallthat—until recently—no technology could begin to emulateanimal nose. Now, however, we have artificial or “electronnosesf9,10g with substantial and growing capabilities. They nsamplers—sniffers—of comparable aptitude.

The closely related issue of chemical trace sampling haswise received little attention. S. F. Hallowell, head of the U

Contributed by the Fluids Engineering Division for publication in the JOURNAL of

FLUIDS ENGINEERING. Manuscript received by the Fluids Engineering Divis
February 10, 2005. Review conducted by J. Katz.
Journal of Fluids Engineering Copyright © 200

i-g,

onrtty.,te

isnntarehery,he

toen

ne”

e-.

Department of Homeland Security’s Transportation Securitysaid “Chemists have been so fixed on detector developmentfthatgthat’s exactly what we got: very well-developed detectorshave no front ends. We’re going to have to reach out to odisciplines to develop novel sampling systems”f11g. Fluid dy-namics is the principal discipline that must fulfill this need.

1.1 Scope.This is an unusual Freeman Lecture: Rather thdetailed review of a current fluids topic, it attempts to bringgether several diverse fields—some outside the traditionalrealm—in order to introduce a novel biofluid dynamics toRather than the culmination of a career, it is an attempt to sout in a new direction.

Of course, not all chemical sensing methods require fldynamic sniffingper se. An early distinction is made here btween trace and bulk detection, for example. The latter invonon-olfactory methods to sense significant quantities of a tmaterial such as a concealed explosive or other contraband.direct-contact methods like swabbing can acquire trace sawithout an overt fluid-dynamic step.

A similar distinction is made between standoff and point detion f12g. Standoff detection requires physical separation betwthe sensor and the site of interest. In this paper, except foscenting and chemical plume tracing, sniffers are consideredpoint-detection devices.

The well-developed fields—some broader than others—bear on the present topic include:

1. Biofluid dynamicsf7,8g.2. Animal olfaction, neurophysiology, and evolution.3. Artificial olfaction, often called the “electronic nose.”4. Airborne particle sampling of the sort used to determin

quality f13g. sIn f14g more than 1000 such instrumemanufactured by 240 companies were identified.d

5. Inhalation toxicology, where laboratory animals are used to

MARCH 2005, Vol. 127 / 1895 by ASME

tur

r o

ofeir

nti

ly-araeneyoalistity.ctioroleionts a

o

eretith

es.p

hy

m-igh-oventp

las

ithofontsr

y oef

afro

c-

ions itoeent

flo-

ex-

,orcs

par-ieren

tiontionand

nts. It

es—

assess human responses to inhaled pollutantsf15,16g.6. Ventilation, especially inlets and hoods for industrial cap

and local exhaust of fumesf17–20g.7. Electronics cooling, where high convective heat transfe

curs in confined spacesf21,22g.8. Aerobiologyf23–25g, “the science of the aerial transport

microorganisms…together with their transfer to the air, thdeposition, and the ensuing consequences…” C. S. Coxf23g.

9. Analytical chemistry, the science of identifying and quafying compounds using laboratory instruments, e.g.,f26g.

1.2 Goals.Lighthill, in the introduction to hisMathematicaBiofluiddynamicsf7g, wrote “The value of seeing any biofluiddnamic problem…against the background of a systematic comptive survey of the biological function in question in many differgroups of animals, can hardly be overestimated.” Such a survsniffers is the principal aim of this paper. It is intended for a braudience, including zoologists, biologists, environmentaanatomists, and physiologists as well as the fluids communalso aims to encourage the fluids community to consider olfaand the olfaction community to include fluid dynamics. Theof sniffers is examined in nature, instruments, and applicatthe state-of-the-art is summarized, and future developmenprojected. It is, in other words, a fluids engineer’s viewolfaction.

2 Fluid Dynamics of Sniffing and SamplingA brief overview of some pertinent fluid dynamics is given h

as a resource for nonfluids-oriented readers. This is no subsfor a basic fluid mechanics course, but it does include sometory and a collection of pertinent—if little-known—fluids issuThe breadth of the topic and that of the anticipated audiencecludes mathematical rigor in favor of an approach based on pcal reasoning, phenomenology, similarity, and scaling.

2.1 Flow Visualization.As elegantly stated by Klinef27g andRoshkof28g, solutions of the equations of fluid motion have liited value without some accompanying phenomenological insusually gained from flow visualizationf29g. This marriage of fluids, optics, and artistry serves research for purposes of discexploration, illustration, qualitative insight, and nowadays quatative measurement. Modern flow visualization includes a comtational element and benefits from both computer andtechnology.

The visualization of olfactory flows apparently began wPaulsenf30g, who placed litmus paper inside the nasal cavityhuman cadaver head and caused the head to “inhale” ammThis early surface-flow visualization method revealed elementhe internal airflow pattern. In the same era, Zwaardemakef2gobserved the flow of breath from the human nostril by wamoisture condensation on a cold mirror. This method is still usin modern times,f31g and Fig. 1sad.

More modern attempts to visualize the airflow in the humnasal cavity used transparent plastic models ranging in scale1:1 f32g to 20:1f33g. Computational fluid dynamicssCFDd simu-lations, e.g.,f16,34g, agreed with experiment in that only a fration of the inspired airflow reaches the olfactory region.

Brueggemann and Jeckstadt also reached that conclus1938f35g following chemical tracer and dust deposition studiethe nasal cavity of a dog. Dawesf36g used cigarette smokevisualize the airflow through a thin slice of a dog’s nose betwclear plastic plates. Such particle visualization is also importaexternal airflow studies of chemical plume tracingf37,38g andventilation exhaust effectivenessf39g.

Underwater, Teichmann used dye to study the waterthrough the olfactory lamellae of an eelf40g. Similar visualiza
tions of both the antennule flows of crustaceans and the chemi
190 / Vol. 127, MARCH 2005

e

c-

-

-tofds,Itn

s,re

f

uteis-

re-si-

t,

ry,i-u-er

aia.

of

ful

nm

inn

nin

w

plumes they detect are shown inf41–43g. Invoking fluid-dynamicsimilarity, humanf44,45g and lab-rat nasal flow patterns wereamined in water models with dye injection.

In contrast, the optical flow visualization methodssschlierenshadowgraph, and interferometryf29,46gd were used only once folfactory research,f47g and Fig. 1scd. This requires special optibut has the advantage of imaging air currents without tracerticles. A National Research Council study recommends Schlimaging for the detection of explosive vapor plumesf48g.

In summary, there is a wealth of resources on flow visualizaf29g that can be used in future olfactory research. Visualizashould always be done first; it defines the flow phenomenasets the stage for more quantitative and detailed measuremeteaches you the physics of the flow.

2.2 Definitions and Assumptions.Sniffing is sampling thsurrounding fluid by olfaction, and a sniffer is the apparatu

Fig. 1 Examples of olfactory flow visualization in nature. „a…condensed moisture traces of a rat sniffing on a Zwaardemakermirror, a form of surface flow visualization, courtesy F. Bojsen-Møller, and „b… tracer particles disturbed by a dog sniffing ahorizontal surface, and „c… schlieren image of the exhalationfrom a dog’s nose †47‡

calwhether natural or artificial—that sniffs. A sniffer has an external

Transactions of the ASME

rsothes t

s ths. E.omemonve.ms ndr

ctiohoigibcaenreensumnsth

s is

arec

eqpuflowflowe-uid

l-flor alet, a

nth

snis

o

seoseca

o

ationsdequa-

is

sideording

singdden

q.ts or

s theastesss

thee

load

terro-f the

aledam-e

gpling

ion,nalresentby

en-

ordseffi-

th by

rface

thise at

vey ita pre-tlyhavensorygant

rhemi-withe ar-pes

naris, or nostril, and a nasal cavity containing olfactory sensovertebrates and in artificial nosessthough the latter do not alsbreathe.d Invertebrates have their olfactory sensors outsidebodies, but we consider them to be sniffers as well. Sometimesensor is stationary with respect to a moving fluid, other timesniffer transports fluid through a stationary sensor apparatuther way, mass transfer by fluid flow is required for olfaction

Olfaction is the sensory detection of an odor or scent, schemical signal in the environment. The sources of such chcals need not be present for olfaction to occur. Moreover,trace amounts, not bulk amounts, are required. A trace is asmall chemical signal—sometimes only a few molecules. WMcGann has defined trace detection as “the process of sacollection, detection, and identification of targeted substancemeasurable by any other means.” For present purposes weno distinction between trace detection thus defined and olfa

Incompressible single-phase fluid flow is assumed througthis paper. Solid particles, when present, represent a neglmass fraction. Weak odorant concentrations are passive sthat do not change the gas properties. Reactions between sand odorants do not alter the energy level of the flow. Thdimensionals3Dd unsteady flow occurs in the most general sealthough lower dimensionality and quasisteady flow are assin many practical examples. The flow may be laminar, trational, or turbulent, depending upon a Reynolds numberranges broadly over the phenomena of interest. Viscosity inored in external flows away from surfaces. For internal flowlumped friction loss is assumed for simplicity. All body forcesneglected. Readers unfamiliar with these assumptions shouldsult a basic fluid mechanics text, e.g.,f49,50g.

The equations to be solved, then, are the Navier-Stokestions of continuum fluid motion, generally by numerical comtation and experimental analog. Often, though, the externalof sniffers lend themselves to simpler solutions like potential-approximations, where fluid rotation is neglected. Softwaravailable to do these simple solutions, e.g.,f51g, and some examples are shown later. The underlying theory is covered in fltexts and does not bear repeating here. However, potentiatheory is inadequate for the flow inside an animal’s nose oartificial nose, where friction is important. Several compNavier-Stokes solutions of such nasal flows in rats, monkeyspeople were done by Subramaniam et al.f16g, Kimbell et al.f52,53g, Kepler et al.f54g, and Kimbell and Subramaniamf55g.

2.3 Modeling and Scaling.A simple modeling exercise cahelp to introduce the key parameters of sniffing. Considerrudimentary sniffer sampling the atmosphere in Fig. 2.sMuch ofthis paper concerns ingenious ways of sample acquisition bying, the simplest of which is shown in this figure.d A certain masmo of odorant is dispersed in the air within an odor cloudvolume Vo, yielding an average odorant mass concentrationCo=mo/Vo. A sampler, or nose, approaches the vapor cloud cloenough that much of the cloud lies within the “reach” of the ni.e., the maximum distance over which it can induce a signifiairflow. By inhaling a volumeVi through the nostril at① overtime intervalDt—“sniffing”—the nose transfers odorant fromVo

Fig. 2 Diagram of a basic sniffing process „the actual interiorof a dog’s nose is much more complicated than this …

to its internal sensor chamber, which has a volumeVsc, at a flow

Journal of Fluids Engineering

in

irheei-

ei-

lyryJ.pleotawn.utle

larssors-

e,edi-atg-a

on-

ua--s

is

swn

end

e

ff-

f

ly,

nt

rateQ=Vi /Dt.The sniff is caused by a reduction in pressurep2 compared t

ambient pressurep1=pa. The pressure differentialDp=p1−p2 in-duces the desired airflow. Conservation of mass between st① and② requires thatU1/U2=A1/A2, whereU is the airspeed anA is the duct cross-sectional area. The steady-flow energytion between stations① and②, ignoring any heat transfer, is

p1

r+

U12

2=

p2

r+

U22

2+ gshf + Shmd s1d

whereg is gravitational acceleration andr is the gas density. Thfamiliar relation reveals that wall-friction lossesghf and “minor”lossesgShm occur at the expense of the fluid kinetic energy inthe nose, and are balanced by the pressure differential accto

Dp = p1 − p2 =rKsU2

2 − U12d

2s2d

whereK is a lumped nondimensional loss coefficient expresthe sum of wall friction plus any duct losses caused by suexpansions, contractions, bends, etc.K is not knowna priori for acomplicated nasal passage, but it can be evaluated from Es2dusing basic pressure and velocity data from experimencomputations.

This textbook material stresses to artificial nose designerimportance of minimizing the minor losses. Each of these wa portion of the velocity head 1/2rU2, and such flow energy lomust be made up by the lung or blower or pump that drivessniffing processssee Sec. 5.3.2d. Narrow flow constrictions aralso lossyf45g. Inside an animal, too many minor losses overthe lungs and impede olfaction, if not respiration itselff56g. In anartificial nose, after the airstream has been sampled and ingated, adiffuser is needed to recapture the potential energy oflow before it is discharged to the atmosphere.

Returning to Fig. 2, if no extraneous unscented air is inhandVscøVo, then the odorant concentration in the sensor chber is simplyCsc=Co. Of course this is unlikely, for in reality thnose also inhales some extraneous ambient airVa, thus reducinthe concentration of odorant in the sensor chamber by a samefficiency factorhs,Va/Vi. The sensed odorant concentratthen, isCsc=hsCo. Other inefficiencies can occur, such as sigloss to the walls of the nasal passages. When particles are pin the odorant cloud, they may also be lost during sniffingsettling or impactionf57g. Realistically, then, the odorant conctration Csc in the sensor chamber is often much less thanCo,which itself is usually tenuous in the environment. In other wthere is an “impedance mismatch” due to poor samplingciency, by analogy with the transfer of electrical energy.

Large impedance mismatches in sniffers are often dealt wipreconcentration, in which a fractionms of the odorant massmo issampled, usually by being adsorbed or impacted upon a suduring the sniff, and the large original volume of airVi is dis-carded.sThe additional apparatus and ductwork required to doare not shown in Fig. 2.d After the capture, artificial noses usmuch smaller volume of carrier gasVp to collect the odoranmass, which is thermally desorbed from the surface, and conto the sensor chamber. This process takes time, but yieldsconcentration factorhp=Vi /Vp as large as 1000 or more, greaameliorating the impedance mismatch. Natural nosesevolved the ability to sense trace odorants directly via the setissue upon which they collect: a much-faster and more eleapproach.

Having arrived at an odorant concentrationCsc in the sensochamber by sniffing, noses sense the odorant by way of a ccal reaction. In nature, odorant molecules interact directlyreceptor cells mediated by a mucous layer, for example. Trutificial noses mimic this, while other man-made detector ty
usually interrogate the captured odorant by spectroscopy. The
MARCH 2005, Vol. 127 / 191

ysiis

inity?ra

y r

ar a

oet

of” in

of

rth

h.5th

lraneth

d,

buThp-

in-

n,intxameidef th

plneonlallyith

scen

eeg,

herof t-ctiohass

ae

arpo.

ionp-

t en-ntila-out toape.e a

sink

outof an

d inatysi-

theets:n airair.h of

o-

for

xis,dis-ulente

natural response to odorant concentration is the psychophpower lawsStevens’ lawd R,Csc

n , wheren is less than 1 anddependent upon the particular odorant speciesf58,59g.

Now from a different viewpoint, the total odorant massmssampled during the sniff is given by

ms = hs ·Q ·Co · Dt s3d

This shows that higher sniffing flow rates and longer sniffscreasems. But does that mean greater olfactory sensitivStuiver f44g reasoned that a higher flow rate reduces ododwell time upon the receptors and thus elicits less olfactorsponse. Mozell et al. and Hahn, Scherer, and Mozellf58,60g fur-ther noted that increasingQ by raisingDp with fixed Dt in olfac-tion studies also increasesVi and changes several variablesonce, causing confusion. The design of the sensor chambebears upon this questionsdiscussed in Sec. 5.3d.

A longer sniff at the same flow rate, however, clearly dexpose the receptors to more odorant mass according tosimple quasisteady-state modelswhich neglectsadaptation, theloss of sensor response with timed. Dogs can do this in caseweak or inaccessible scents by taking 1/2 Hz “long sniffsplace of their normal,5 Hz sniff ratef47,61–63g.

Finally, the Reynolds number Red=Ud/n governs the naturethis flow, whered is the nasal passage diameter andn is thekinematic viscosity of the air. Below Red,2000–3000, laminaflow is expected in simple ducts, though not necessarily incomplicated nasal passages ofmacrosmaticanimals—those witkeen olfactory powers. Sincen is roughly constant at about 1310−5 m2/s for air in the near-ambient temperature range,Reynolds number varies mainly with airflow rateQ and physicascaled. The anterior nasal passages of terrestrial vertebratesfrom about 1 mm diameter in the ratf64g to a centimeter or morin the largest animals. Given rat respiratory flow rates in100 ml/min–900 ml/min range, a typical Red is about 350 anlaminar nasal flow is assumedf53g, at least initially. In humanshowever, the issue is controversialf45g.

Underwater,n is an order of magnitude smaller than in airthe pertinent length scale is also often much reduced.chemosensory hairssaesthetascsd of crustaceans, for example, oerate with Red,1, where the interplay of fluid viscosity andertia is delicately balanced,f65g and Chap. 15 off66g.

2.4 Potential-Flow Inlets.The potential flow approximatiomentioned earlier, naturally describes the flow of ideal fluidssuction orifices like the nostrils of present concern. We can eine external sniffing airflows by simple means, so long as Rlarge and both wall friction and flow separation are either avoor ignored. Ventilation engineering takes major advantage oas is detailed in several referencesf17–19,67,68g.

In brief summary, potential flow theory reveals that a siminlet is not directional. Wile E. Coyote’s ACME vacuum cleamay suck in all sorts of objects from a distance, but thatworks in the cartoon world. In reality an inlet draws fluid equfrom all available directions, so its influence drops rapidly wdistance. Consequently, except for the case of airborneplumes, sniffing is not a stand-off activity and proximity is esstial for a nose to acquire a localized scent.

Potential flow also explains the inefficiency of a simple frstanding pipe as an air inlet: Even when aimed at somethindraws air equally from behind as well as from in front. It furthas a high loss coefficient, 0.93, meaning that it wastes 93%velocity head 1/2rU2. A flanged inlet, Fig. 3sad is an improvement: The forward reach is extended by excluding any sufrom the rear hemisphere. Still, the sharp-edged inlet orificeloss coefficient of 0.49. Whether a nose is powered by lungbatteries, such an unnecessary pressure loss is a burden.

Nature abhors sharp edges in favor of smooth, bulbous,dynamic nostril inlets like the one shown in Fig. 3sbd. Also called
a bellmouth, this inlet has a very small loss coefficient. Here is a
192 / Vol. 127, MARCH 2005

cal

-

nte-

tlso

shis

e

e

ge

e

te

o-

isdis

ery

ent-

-it

he

na

or

ro-

important design principle for artificial noses: avoid the shedge and the step change in area. It is the natural thing to d

A definition of the “reach” of an inlet is the size of the regupwind of the inlet from which all of the fluid is ultimately catured f19g. In Fig. 2, in order for the sampling efficiencyhs toapproach unity, the reach of the sniffer shown must at leascompass the entire odorant cloud. Industrial local-exhaust vetion has the same problem: The exhaust hood must reachcollect fumes from welding, for example, and not let any escGiven the limited reach of a potential-flow inlet, this is quitchallenge. For the case of Fig. 2, approximated as a simpleflow, the induced airspeedu=Q/pr2 drops linearly with distancerforward of the nostril inlet.

Potential-flow inlets scale up or down geometrically withregard for nondimensional numbers like Re, so the reachinlet also grows linearly with its diameterd because Q=pd2U1/4. This works against the small-diameter tubing usewand-type leak samplers, e.g.,f69g. They have little reach eventhe extremeU1 corresponding to choked flow, and must be phcally inserted into an odorant cloud to sample it.

2.5 Turbulent Jets. There is a key distinction betweenbehavior of potential-flow inlets just described and fluid jwhile inlets are omnidirectional, jets can be vectored. Aim ajet in a certain direction and it maintains that direction in stillThe “reach” of a jet, then, is many times longer than the reacan inlet having the same flow rateQ. Jets thus project fluid mmentum at a distance.

The turbulent jet literature is vast. Useful surveys of itpresent purposes are found in Chap. 4 off17g and inf70–72g. Therecent discovery of synthetic jets is also pertinentf73g.

A jet entrains fluid from all directions perpendicular to its awhich causes it to grow linearly in diameter with increasingtance downstream. The volumetric entrainment rate of turbjets is approximately 0.25x/A1/2, wherex is the distance from th

Fig. 3 Streamlines „arrowed … and equipotential lines „solid …

for „a… a flanged, sharp-edged inlet and „b… a bulbous “natural”bellmouth inlet. „These planar 2D potential flow solutions areshown only for illustration purposes …

nnozzle andA is the nozzle cross-sectional areaf70g.


ae

alreof.3.enila

ec

othlestw

spestigl-

abed

dclol-In

nerg ongticntr

bleen

rvertioe f

onr,

cha

ob

nis is

Ife ofbeif

l inans-epo-

-intooften

ing af thert

n theicklyand

n the

oughans-ed to

d ad-

of theow

rt inber,

ntqual.

l-

uch-this

sion

neby

e ex-y areuilib-

alto-l and

c.

wdialeldowsugt’s

man-ciple

strilese.

When a jet impinges upon a solid surface it tends to attachto exhibit starting impinging jet behaviorf74g. Free jets impingupon surfaces to become wall jetsf75g.

Even though jets seem at first antithetical to the present gosniffing, they can be useful on at least two counts. First thejet-assisted olfaction, where jets help to focus the suctionsniffer in the forward direction to improve its reach, Secs. 4and 5.2.4. Second is the use of jet momentum to stir up theronment, resuspending settled particles and making them avafor olfaction as shown in Fig. 1sbd.

Phares, Smedley, and Flagan at Caltech have studied the stopic at lengthf76g. They explored the effects of Red, nozzleheight, and jet velocity profile on particle removal using bnormalf77g and obliquef78g jets, and found that surface particrespond to the shear stress imposed upon them by jet impacf79g.Particle removal depends on particle size and jet duration asf80g. With these results in hand, they examined previous resusion theories and proposed a new model. Other recent invetors f81g also studied this topic, and onef82g suggested an anaogy with heat transfer enhancement by pulsed jets.

In applied experiments in the Penn State Gas Dynamics Lratory f83g, short-duration air jets from round nozzles impingupon human subjects whose clothing was contaminate10-micron-range trace particles of the explosive cytrimethylene-trinitraminesRDXd. Dislodged particles were colectedsSec. 6.3.2d and quantified with results shown in Fig. 4.addition to the shear-stress mechanism described above, idetachment of particles also occurs due to jet impact rufflinthe clothing. No matter what the mechanism, though, jet impiment has become the principal means of liberating trace parfrom people for analysis and detection in a reasonably nonisive manner.

The quasilinear response shown in Fig. 4 is understandathat the nozzle-to-person distance was fixed in the experimand the nozzle exit was sonic. Since jet momentum is consean increase in stagnation pressure thus produces a propoincrease in jet momentum, which is thought to be responsiblinertial particle removal from clothing.

2.6 Particles and Stokes Number.Can particles in a flowfollow streamlines, as molecules do by definition? This questianswered by considering the nondimensional Stokes numberatio of the aerodynamic response time of a particle to theacteristic flow time,

Stk=rp ·dp

2 ·U

18 ·m f ·Lcs4d

whererp is particle density,dp is diameter,m f is fluid viscosity,and Lc is a characteristic length, such as the size of a flow

Fig. 4 Portal results of liberated RDX particle mass as a func-tion of impinging-jet stagnation pressure †83‡

struction. AsStkapproaches 0, particles have no significant inerti


nd

ofisa1vi-ble

ond

elln-a-

o-

by-

tialf

e-lesu-

intsd,nalor

isther-

-

and they faithfully follow streamlines, but forStk greater thaabout 1, impaction of particles upon obstructions is likely. Ththe basis of devices calledimpactorsthat collect particles fromflowing fluids.

Consider airborne particles of unit specific gravity and 10mmdiameter, having masses of about a nanogram each.U=30 m/s andLc=1 m, roughly the speed of a car and the sizits windshield, thenStk=0.009 and the subject particles willlikely to flow around the windshield without striking it. ButLc=5 mm, the diameter of the radio antenna, thenStk=1.8 andimpaction is much more likely. Calculations like this are usefuthe design of particle collectors, the loss of particles during trport in tubes, particle settling time, terminal speed, and the dsition of inhaled particles in the nose and lungsf57,84–86g.

2.7 Advection Versus Diffusion.Advection is bulk fluid motion whereas diffusion is the molecular spreading of one fluidanother without bulk motion. These distinct phenomena areconfused in the olfaction literature. In Feynman’sf87g classicclassroom example gone wrong, diffusion is taught by openbottle of ammonia and waiting for the student in the back oroom to say “I smell it.” But if diffusion is the only transpomechanism, he will not smell it by the end of the class, or eveend of the semester. In reality, the ammonia molecules are qutransported by the advection of air currents around the roomby thermal convection currents generated by the people iroom.

This point is often made but the confusion persists, even thit may be just an imprecise choice of terms. Airborne odor trport in the nasal cavities of vertebrates has been attributdiffusion f3,88,89g, but that too is unlikely.

Vogel draws an elegant comparison between diffusion anvection for a popular audiencesf90g, p. 182d. Diffusion onlyworks over short distances and thus depends on the sizeflow in question, “But even in air, diffusion remains glacially slfor what we regard as ordinary distances.”

The relative rates of advective versus diffusional transpofluids are described by the nondimensional Péclet numPé=LcU /D, where D is the diffusion coefficient of an odoramolecule in the fluid of interest. When Pé=1, the rates are eReturning to the rat’s-nose example of Sec. 2.3, letLc=1 mm andU,5 m/s f53g. Diffusion coefficients of typical odorant moecules in air are broadly in the range of 0.1–0.3 cm2/s f91g, yield-ing a rat’s-nose Péclet number of several hundred. Only at mlower airspeeds and length scales can diffusion matter insniffing example. But there are certainly cases where diffumatters in olfactory sensing, such as lobster “sniffing”f92,93g.

Griffy f94g calculated the time required for a trinitro-toluesTNTd bomb to produce a saturated TNT vapor level in a roompure diffusion: hundreds of days. He concluded that airbornplosive vapor detectors are unlikely to succeed unless theorders of magnitude more sensitive than needed to detect eqrium vapor levels. But let us not abandon vapor detectiongether, so long as there are air currents to convey the signasniffers poking their noses into nooks and cranniesssee also Se5.2.2d.

2.8 Vortex Flow. The potential-flow vortex is a rotating flowith a strength determined by its core singularity. It has no ravelocity, circular streamlines, and a circumferential velocity fithat dies off as the inverse of radius from the core. Vortex flare covered in elementary fluids textbooks, but see also Lscholarly bookf95g that appeals to a broad audience.

What relevance do vortices have to sniffing? At least onemade sniffer uses an apparent vortex as its operating prinf96g. The puffs on either side of the dog’s nose in Fig. 1sbd are thestarting vortices of impinging nostril jets. The bellmouth noflowfield of Fig. 3sbd is produced by twin counter-rotating vorticshidden inside the nostrilsd, and so on. Vortices are everywher
a One vortex of special interest is theintakeor inlet vortex. There
MARCH 2005, Vol. 127 / 193

sucsat

thats

romth

uu-duci

lemaal

ds fc-ticda

o tuhad

wsar

ntsseth

hawsvieng

tubllat-t Ath

y oct

f air-Out-windaxisn betion.

iters/ftop

tm in

ifytion

e im-eter

esti-tickying

wide

rates

ns-tual

o bee-e at-esentifyple

, pu-on is

d-uffi-anunder

oth-gas,

edlogy

oadpeldoor

er?re isvityf the

t inafter

mis-ll but

thek

drivesand

alogy

note, butother

are photos of vortex cores extending from the idling jet engineparked aircraft, touching the ground like mini tornadoes and sing up debris. It is a common fallacy that the jet engine createown inlet vortexf97g, but researchers in E. M. Greitzer’s labMassachusetts Institute of Technology have demonstratedcross breeze is also requiredf98g in all cases except those of inlewith swirl vanesf99g. The cross breeze sheds a line vortex fthe engine inlet lip that turns downward and attaches toground.

The inlet vortex can be duplicated on a tabletop using a vaccleaner and fan according to Stongf100g. This instructive demonstration is not difficultsFig. 5d, but its stability is greatly improveif one places a screen or flow straightener after the fan prodthe cross breeze.

2.9 Frequency of Animal Sniffing. Although dependabfluid-dynamic data on olfaction are rare, enough exist to estithe nondimensional sniffing frequency of at least a few animThis is the Strouhal number, Str=fd/U. For the rat, dog, anhuman, respectively, the available data are approximately alows: dimensional sniffing frequencyf =10, 5, and 2 Hz, charateristic nostril dimensionsd=1, 7, and 13 mm, and characterisairspeedU=8, 5, and 3 m/sf33,47,53,89g. These values yielStrouhal numbers of about 0.008 for the dog and human0.0014 for the rat. Thus sniffing is a slow process compared tflapping of animal appendages for propulsion, where Stronumbers are up to two orders of magnitude larger. A quasisteflow approximation is justified in the analysis of sniffing airflof33g at this low frequency. It may be that animal sniffing ratesbased, at least in part, on the need to provide sharp gradieodorant rather than a constant odorant level to the olfactorysors. Adaptation works against high sensitivity in the limit assniffing rates approaches zero.

3 Traditional Sampling Issues and MethodsSampling the environment is important to society. So much

been written on this topic, including several thorough reviethat it serves the present purpose only to give a brief overwith some key references, noting the issues relevant to sniffi

3.1 Airborne Particle Sampling. The U.S. Clean Air Accontrols particles in the atmosphere in order to protect phealth and prevent environmental damage. Airborne particuin the,10 mm sPM10d and,2.5 mm sPM2.5d aerodynamic diameter range are regulated by law according to National AmbienQuality Standards. Compliance is monitored by samplersdraw in ambient air and collect the airborne particles by anseveral means, including impactors, cyclones, filters, and elestatic precipitatorsf13,25,101,102g.

Fig. 5 Inlet vortex into a vacuum cleaner hose with a crossbreeze, visualized by coating the ground plane with talcumpowder

The aspiration efficiency of these sampling inlets is one con

194 / Vol. 127, MARCH 2005

ofk-its

t a

e

m

ng

tes.

ol-

ndhealy-

eofn-e

s,w.

ices

iratf

ro-

cern, since they must sample a representative population oborne particles. Isokinetic sampling helps to accomplish this.door samplers also require inlets that are insensitive to thedirection. Rooftop inlets are often symmetric about a verticalfor this reason, while thin-walled sharp-edged inlet tubes cashrouded to make them less sensitive to variable wind direcThere are many commercial samplers available, from the lmin range up to about a 19 liters/s flow rate for large roomodelsf84g.

Aside from filtration f101g, impaction sSec. 2.5d is the moscommon means of separating the particles from the airstreathese samplersf13,84,102g. Multistage impactors further classparticles into several size ranges. This matters in air pollumeasurement but is not an issue in olfaction; a single-stagpactor can sample all the particles above a minimum diamtogether if purely for chemical detection.

A dog’s nose collects particles by impaction in the nasal vbule region, where the respiratory mucosa are moist and sf103g. Preventing impacted particles from bouncing or blowoff is likewise an issue in man-made impactors, where avariety of sticky coatings is usedf13,84,102,104g.

Another approach is the virtual impactor, which preconcentparticles in the sampled airstream by a factor of 6:1, 20:1f84g oreven 25:1f105g. The particles remain airborne, which helps traport them to an analyzing instrument. Design rules for virimpactors are given inf84g.

Other than inert particles, there are also bioaerosols tsampledf12,25,106–108g. With such a bewildering array of spcies already present in the air, detecting a biological warfartack is a problemf109g. The sampling step must not harm thlive airborne particles, since a culture is often required to idethem f110g. The analysis step may also require a liquid samlike that provided by impingers and some cyclone samplersrification may be needed, and a high sample concentraticalled for f12g.

3.2 Indoor Sampling. Indoor air quality considers the averse health effects of smoking, sick building syndrome, inscient ventilation, etc.f111–113g. Airborne sampling indoors begwith the realization that the dangerous pollution sources areour noses, including chlorinated water, dry-cleaned clothes, mballs, air fresheners, cooking, paint strippers, solvents, radonand especially tobacco smokef112g. Large government-sponsorstudies like Particle Total Exposure Assessment MethodosPTEAMd f114–116g in the United States and others abrshowed that personal PM10 exposure levels measured by lasamplers are significantly higher than either outdoor or inlevels measured by fixed samplers.

3.2.1 Personal activity cloud or human boundary layWithout questioning the results or significance of PTEAM, hean alternative fluid-dynamic interpretation. A “personal acticloud” was proposed by PTEAM investigators as the source omeasured discrepancy in exposure levelsf116,117g. Wallacef118gexplains: “It is almost as if the participants were walking aboutheir own personal cloud of particles, a sort of Pigpen effect,the character in thesCharles Schulzd Peanutscomic strip.”

The cloud analogy and the comic-strip reference are sadlyleading, though, for what actually happens is not a cloud at arather a rising thermal boundary layer that separates to formhuman thermal plumesFig. 6sad f46gd. It was first shown by Clarand Coxf24g, Lewis et al.f119g, and Clark and Edholmf120g thatthe temperature difference between people and ambient aira natural-convection boundary layer beginning at the feetseparating from the head and shoulders. Thus the “cloud” anoverlooks the strong vertical transportsU,0.25 m/s,Q up to50 liters/sf121gd in the airflow about a person. This transportonly presents floor-level contaminants to one’s breathing zonalso entrains the surrounding air and its particle burden at all
-levels below the breathing zone. Quotingf122g, “The natural con-

lespo

reiontiv

va-e.da

co

pendsk

inskno

ve

drtinys t

is islaterbe

thetiveso

pen-r lo-mplerwardmingsed toiding

eto betud-ents.the

, im-the

mplee im-or-h a

storer skinef-g the

ile,plings arepoly-g thectedeter

sess aay to

usesev-

enttions in-of

viron-tabledtooex-

m-emi-

sim-ling

’sf

vection boundary layer around the human body is capabtransporting particles such as dust, skin scales, pollens, andand provides a link in the chain of airborne infection.”

A second problem with the PTEAM interpretation is its disgard for the role of human skin flakes: “body cloud emissare…not considered here as a component of the personal accloud effect” f117g. In fact, human skin is easily the most prelent particulate in the human thermal boundary layer and plumcomplete layer of human skin is desquamated every 1–2f24g, releasing a million skin scales/min with a 14mm averagediameter and a size range of 5–50mm (see Fig. 6sbd f123,124g).Most inhaled air comes from the human boundary layer thattains these particles, from which 6000 to 50,000 5–50mmparticles/liter of air enter the human nose. Most clothing ismeable to this particle stream. Ordinary house dust is found umicroscopic examination to consist of 70%–90% humanflakes covered with microorganisms. According to Syrotuckf125g,if you walk at 1 1/3 m/ss3 mphd you leave behind 500 skflakes/m. The weight of this evidence on the significance offlakes in the human microenvironment ought to be hard to igssee alsof125–128gd.

Moreover, human skin flakes contain mitochondrial DNA, ethough they have no cell nuclei and thus no nuclear DNAf129g.We thus continuously shed gross samples of our mitochonDNA into the surrounding air. Anthropologists sequencingremnants of ancient mitochondrial DNA must be scrupulou

Fig. 6 „a… Schlieren image of the rising boundary layer andthermal plume from a human being „L. J. Dodson … †46‡ and „b…scanning electron microscopy image of a desquamated humanskin flake, H. A. Gowadia †123,124‡. According to Syrotuck†125‡, “They are cornflake in shape which gives them an aero-dynamic characteristic.”

avoid contamination from their own airborne skin. We cannot sup


ofres

-sity

Ays

n-

r-er

in

inre

n

ial

o

press this shedding of our DNA, and collecting it for analyscertainly possible according to principles to be describedsSec 6.3d. Ethical questions about this possibility remain toresolved, however.

Meanwhile note that a thorough study of the airflow inhuman breathing zone, including flow visualization, quantitameasurements, and CFDf130g has apparently not been done,there is much still to learn about the reach of inhalation, theetration of exhaled nostril jets, particle intake, and the propecation of personal samplers. Where to locate a personal saon a breathing person, considering the nose inlet flow and upboundary layer motion, still remains an open question. In seeproof that nothing is sacred, even intranasal samplers are usample incoming particles inside human nostrils, thus avothis pitfall f131g.

3.2.2 Personal samplers. Referencesf14,15g describe thstate-of-the-art of small battery-powered personal samplersworn by human subjects for particle exposure monitoring in sies like those just described and in industrial hygiene assessmTypically a battery-powered diaphragm pump draws insampled air at a few liters/min through a sorbent tube, filterpactor, or cyclone. The inlet is usually connected by tubing topump, and is attached to a subject’s collar or lapel to sabreathing-zone air. A miniaturized five-stage personal cascadpactor is availablef132g to classify particle sizes. The SKC Cporation’s Button Sampler is a collar-clip filter sampler witporous curved-surface inlet having low wind sensitivityf133g.

Mannequins are used to test personal samplersf134,135g, butthey do not always properly simulate people. A department-mannequin is the worst case, for it has neither body heat noflakes and it cannot breathef136g. Human subjects are thus prerable to mannequins here and in any experiments involvinhuman thermal boundary layer and plume.

3.2.3 Hand vacuums. Hand vacuums are simple, versatfilter-type “dustbuster” sniffers. They see regular use in samthe clothing and luggage of aircraft passengers. The filtermade of cellulose fibers, glass or quartz fibers, membranes,carbonate pore material, or plastic foam. In security screeninfilter is removed and heated in a desorber, driving off colletraces to be detected, usually, by an ion mobility spectromf137g.

3.2.4 Wand-type sniffer probes. Wand-type sniffer probewere singled out in Sec. 2.4 for their short reach. Neverthelsimple probe or wand at the end of a hose is a popular wpoke around and sniff for something. Leak detectors oftenwands for gas collection and they are considered prior art ineral inventions, e.g.,f138g. A search of the technical and patliterature for the term “sniffers” mostly yields such leak-detecequipment with long hoses and hand-held, pointed wandtended to pinpoint leaksf69,139g. In one case a filter at the enda long hand-held hose samples particles in clean room enmentsf140g, while in another a heated probe tube feeds a porgas sampling systemf141g. A person probing with a sniffer wanis like an elephant exploring with its trunk: Both animals arebulky to get into tight places without the aid of an olfactorytension tube.

3.3 Outdoor Sampling.Sniffing outdoors raises further coplicating issues having to do with flight, the weather, and chcal plumes.

3.3.1 Sampling by flight vehicles. Sampling by flight vehicleinvolves a mobile sniffer moving rapidly through a relativelymobile atmosphere. Man-made flight platforms for air samprange from miniature unmanned aerial vehiclessUAVsd f142–145gto full-sized aircraftf146–148g. Beginning in the 1970s, NASA5.5 m-wingspan Mini-Sniffersf149g pioneered UAV sampling o
-the atmosphere at high altitudes. Since then the potential of UAVs
MARCH 2005, Vol. 127 / 195

atiim

anrom-ar

etsed

ghfinet

in treat

paeth

e a

berticr aere

ceudesteoue

issanher

log,ces

nsfing

t ofrsialoilals,the

ose as

rmalngreions.

theapor

mi-nt de-rt air.

des-w thattge of

lrome-n be

. A

for biological agent detection has been exploredf144,145g, and arecent Defense Advanced Research Projects AdministrsDARPAd program spawned microair vehicles that could be slarly employed for sniffingf142,150g.

In contrast, take the current Intercontinental Chemical Trport Experiment-North America as an example of sampling ffull-scale aircraftf151g. Its principal flight vehicle, the NASADryden DC-8, bristles with sampling probes of all sorts. Fiveshown in Fig. 7f146,147g. Most of these feature shrouded inlto minimize flow angularity effects, after Kiel’s original shroudPitot probe from the early days of aeronauticsf152g.

In principle, it appears easy enough to extend a tube throuairplane window and sample the air. Some further design rements are obvious, like mounting the probe on a strut to goutside the airplane’s boundary layer. For chemical speciesupper atmosphere, though, there is the concern that they willwith the walls of the tube. Particle sampling is an even greconcern: Recalling the Stokes number from Sec. 2.5, whenU ishundreds of m/s it becomes hard to prevent particles from iming inside the probe rather than being conveyed to the instruminside the aircraft. Likewise the airflow angle with respect toprobe axis is not well known. These issues combine to makborne sampling probe design challenging.

Wind-tunnel testsf153g are often used to verify that new protypes can achieve isokinetic sampling and deliver a true pasample. A good reference summarizing inlet probe design focraft samplers isf154g. The topic is revisited in Sec. 4.3.13, whwe will see how birds solve this problem.

3.3.2 Micrometeorology. In nature, sniffing must take plaoutdoors under all manner of weather conditions. Sun or clocalm, steady, or gusty winds; moisture levels; the temperaturthe air and terrain, all play a role. The broad field of micromerology f155,156g is not reviewed here, rather just a few isspertinent to sniffing.

First, consider the windf157g. The lowest atmospheric levela turbulent planetary boundary layersPBLd that we perceive awind. Meteorological calm, 2.2 m/s or 5 mph, is still a significbreeze for bioaerosolsf107g, insects, and even people, whose tmal plumes become wakes at a much-lower airspeedsSec. 6.3d.Sniffers operate in the smallest local regime of micrometeorothe bottom 2 m or so of the PBL’s roughness layerf156g. Even soill winds readily disrupt sniffing by dispersing chemical tra

Fig. 7 Sampling inlet probes on the NASA-Dryden DC-8, „a…heated, Teflon-lined PANAK probe, „b… U. Hawaii shroudedprobe †147‡, „c… nacelle-mounted ATHOS probe †146‡, „d… wing-tip mounted aerosol scattering spectrometer probes, and „e…shrouded POPS probe „NASA photos …

f158g. Moths, for example, cannot track pheromone plumes whe

196 / Vol. 127, MARCH 2005

oni-

s-

e

ane-itheacter

ct-ntseir-

leir-

s;of-

s

t-

y,

the breeze is too stifff159g. On the other hand wind interactiowith obstacles offer opportunities for olfaction, such as snifthe recirculating flow downstream of a buildingf160g.

Moisture levels are still another matter. Although the effectemperature and humidity on human olfaction is controvef59g, various investigatorsf161,162g show that an increased smoisture level generally aids olfactory trace detection by animwhether of buried food or land mines. Extreme aridity, onother hand, may so dry the mucous membranes of a dog’s nto inhibit olfaction.

Also there is the issue of the buoyant odor-bearing theplume f70,155,163–165g. As an example, consider the followilandmine detection scenariof125,166g: The surface temperatuof the soil varies some 30°C daily, depending on local conditThis desorbs some trace explosives—where present—fromsoil surface over a buried landmine, creating an airborne vsignal. What becomes of this signal, however, depends oncrometeorology. On sunny days a strong temperature gradievelops above the soil, which can be 40°C hotter than deseThis produces unstable thermal convection, transporting anyorbed explosive vapor upward and away. Measurements shotypical thermals arise from surface areas of 1 or more m2, rise aspeeds of about 1/4 m/s, and occur at frequencies in the ran4/min f155g. Visualized thermalssFig. 8 f167gd show typicamushroom-shaped convection cells. Under such adverse micteorological conditions no stable layer of trace explosive caexpected above a buried landmine.

The situation improves, though, from sunset until morning

Fig. 8 Schlieren images of high-Rayleigh-number thermalconvection from a suddenly heated horizontal surface, simulat-ing the earth at sunup on a windless day, courtesy J. C. Mol-lendorf †167‡. „a… Early thermals, „b… a forest of starting thermalplumes develops, with both their crowns and stalks visible, and„c… fully developed thermal convection field. These picturescompliment those in the literature using tracer-particle visual-ization. Here an integrated view is shown, though one canimagine the depth effect

nstable boundary layer often forms above cool soil in the summer.


caurith

alltioartalgn

ans

lan

ntloww

lowo th

icem

n tlu

e Rterndmnts

plihat-en

-s aall

aryrrestes

cue

ctroco

entu

nadithverBhe

e-spoca

-in

gicut

ferse.g.,rn

ade

ylo-ex--

umgy ofr-y

imalre-ago;

s and

rome,nos-

, andsort

ne ofllows-n a

Through the night, explosive vapor from buried landminesaccumulate in this layer. Thus mines are most detectable dthe evening, night, and early morning in calm weather whenground is moist.

Given a prevailing wind, however, the situation is dramaticcomplicated by mixed free and forced convection, both unstaary and fully turbulent. In sunny weather, thermals are shehorizontally and mixed out by turbulence. A light horizonbreeze, less than 1 m/s, is enough to tilt the thermals over sicantly. For higher wind speeds, forced convection dominatesany explosive vapor from a minefield is quickly diluted and traported away in the PBL. Not even trained dogs can locatemines under such adverse weather conditions.

3.3.3 Chemical plume tracing. This naturally leads to a rece“hot” research topic, chemical plume tracing. The ability to foltenuous outdoor plumes in nature rewards many animalsfood or sex. The fluid dynamics of plumes and how to folthem is thus a part of sniffing as we define it here. It was alssubject of a recent DARPA/Office of Naval ResearchsONRd pro-gram that funded multidisciplinary studies to understand chemplumes in nature and to develop artificial plume tracing systMore details are available inf41,92,168,169g.

Briefly, much of the research to date concerned insects iatmosphere and crustaceans in the ocean. The small-scale pthese creatures track are nonetheless large enough that thnolds number dictates fully turbulent flow. They are highly inmittent, having large scales comparable to the plume width acascade of finer-scale eddiesf41,170g. They spread downstreaand contain streamwise and cross-stream chemical gradietheir mean structuref43,171–174g.

An animal’s response to this intermittent stimulus is comcated. Insects and crustaceans have sensory appendages tvide them with plume informationf42,175,176g. Lobsters, for example, flick their antennules to sample the environmf92,93,177g. Both chemical orientationschemotaxisd and flow ori-entationsanemotaxisd are invokedf175,178,179g. Generally upwind progress toward the plume source is modulated by turnsometimes by “casting”f168,180g. Weissburg takes an overview of the fluid dynamics underlying this animal behaviorf170g.

Robots employ such natural plume-tracing principles with ving degrees of success. Man-made chemosensors are cuinferior to natural ones in terms of speedf169g, but neverthelesseveral robotic plume tracers have been developed andf41,181–185g, including the “Robolobster”f186g.

Standoff detection of small localized chemical plumes is arent security concernf48g. Understanding explosive vapor plumdynamics, for example, is helpful in the development of spescopic standoff detectors. Such plumes may have very lowcentrations of chemical vaporssee Sec. 2.6d, but they can bsought and interrogated based on their buoyancy or momewhich drives the trace chemical transport.

These small-scale plumes are in stark contrast to the hugeral and man-made plumes that have grave environmental ancurity implicationsf187g. They are sometimes visibly tagged wparticles, as in Fig. 9, but also sometimes quite invisible. Sehorrific plume accidents now serve as case studies, includingpal and Chernobylf188–190g. Large-scale computational plummodeling, e.g.,f191g, is driven by the knowledge that plumgenerating weapons of mass destruction are within the graterrorists. Plume dispersion in cities, driven by convoluted lmeteorology due to buildings, is especially challengingf192,193g.

4 Nature’s Sniffers and Biomimicry

4.1 Narial Morphology and Evolution. Almost no one except Negusf194g and Bangf195g has shown enough interestthe external animal nares—the nostrils—to do a morpholostudy. It cannot be done justice here for several reasons, b
least one can take a peripheral look and open a prospectus

nnge

yn-ed

ifi-nd-d-

ith

e

als.

hemesey-

-a

in

-pro-

t

nd

-ntly

ted

r-

-n-

m,

tu-se-

alo-

ofl

alat

further study. In that the few fluid-dynamic studies of snifconducted to date have yielded interesting results,f16,32–34,47,65,92,93,176,196g, there is clearly more to leafrom nature’s broad range of sniffing systems. What man-msniffers have to offer thus far is paltry in comparison.

Here we will examine an abridged but representative phgenic cross section of the animal kingdom, with at least oneample from each vertebrate classsmammals, birds, reptiles, amphibians, and fishd, but only examples from the arthropod phylof invertebrates. Special emphasis is given to the morpholomammalian snifferssFig. 10d. Taxonomic and phylogenic infomation is obtained fromf197–200g, and anatomical terminologf201g is adopted except for a few common-usage lapses.

Nature, over some 300 million years, has explored ansniffing systems quite thoroughly. Fossil animal DNA wascently recovered and studied up to about 100,000 yearsenough to address some far-reaching evolutionary questionshake up the old phylogeny in placesf202g. For example, fromfossil mitochondrial DNA and other evidence, theCanidaeappeato have arisen as a distinct family of carnivores s50 million years agof199,200g. Within this family the wolf, dogfox, raccoon, bear, weasel, and jackal have nearly identicaltrils. TheHyaenidaeappeared separately about the same timeare more closely related to theViverridaeandFelidaethan to dogaccording to the DNA recordf200g. Hyenas nonetheless spvery dog-like muzzles and nares.

Despite all her diversity, though, nature never developed oour most clever man-made devices: the turbomachine. Beaction remains the natural way to pump fluidsf203g. On this account a small, low-power pump or fan gives artificial olfactio

Fig. 9 Satellite photo of the ash plume from the eruption ofMount Etna on October 29, 2002. The plume direction is SSEover eastern Sicily, the city of Siracusa, and the MediterraneanSea. The lateral scale is roughly 200 km, and the scale of thelargest visible eddy „i.e., the plume width … is perhaps 10 km.Photo PIA03733 by the NASA GSFC/LaRC/JPL MISR Team

Fig. 10 An abridged phylogeny of mammals for the study ofexternal nares evolution. For brevity the common name isgiven in place of the scientific species name. Time progressesnonlinearly from left to right for compactness, and branchesindicate evolutionary divergences of a group „the lower arm …

forfrom the general mammalian stock

MARCH 2005, Vol. 127 / 197

tins

uid, w

allowu wingmaor-n ofunn renc

omlfax ae

il

andes. A-infr

ltiptiored

nd-nceinc

daFigoseiffinironte

38aai

to t

airiveha

dondesig

fs.ag

eaetl-

l.nts.

ri-y ofntly,ca-cord,ternal

todyair-lyingalarion,

slits

llo-mo-

entralcau-the

y thestril

s aes:ard

tly

aris, is

uringthis

w is

certain advantage over nature’s devices: It can inhale air conously and efficiently, and exhale it somewhere else. Evenwhen it comes to mimicking nature’s olfactory sensors, their fldynamic sophistication and their direct connection to the brainare still far behind.

4.2 Biomimicry. T. H. Huxley said “Sit down before fact aslittle child, be prepared to give up every conceived notion, fohumbly wherever and whatever abysses nature leads, or yolearn nothing.” Biomimicry is innovation inspired by observnature, learning her lessons and deliberately copying them inmade devicesf204g. Here, for example, the study of narial mphology may lead to better sniffers for the new generatioartificial noses now becoming available. This dual need fordamental studies of mammalian olfactory systems has beeognized by funding agencies including the U.S. National SciFoundationf205g.

4.3 Internal-Flow Noses of the Vertebrates.In mammalsand many other vertebrate species, nasal passages lead frexternal naris through a maxillo-turbinate region, past the otory mucosa—nature’s sensor chamber—to the nasopharynthence to the lungsf3,206g. Olfaction is further aided in somspecies by a dedicated olfactomotor systemf63g that drives nostrmotion during sniffing by way of elaborate musculature.

Every species shown in Fig. 10 is macrosmatic except humIn two cases, mammals of the same order are compared in oremphasize diverse narial evolution in closely related speciespecies shown are also of the infraclassEutheriasplacental mammalsd except the opossum, a marsupial, who belongs to theclassMetatheria.

One must beware here, given nature’s penchant for mufunctionality, not to confuse other roles with olfaction. Respirais intimately connected with olfaction in all of the considespecies. Nasal turbinates, for example, are not “turbulators”smix-ing devicesd, but rather air-conditioning devices for heat amoisture exchangef36,206–209g. They provide additional olfactory surface area and generate vorticity even so. Further coarises in using the fossil record to trace olfactory evolution, sfossils do not preserve the soft nasal tissues.

4.3.1 Dog. The dog’s nose is recognized as the gold stanof olfactory acuity. There are many good sniffers shown in10, but dogs are easiest to train for olfactory detection. Our clanimal companions, they are the cheerful butt of endless snhumor because we cannot appreciate the rich olfactory envment as they can. They are the ultimate mobile, instinctive, iligent sniffing platforms.

A brief history of canine olfaction research begins in 19when it was shown experimentally that the majority of inspiredbypasses the olfactory epithelium, which is offset from the mairway of the nasal cavityf35g. Dawes’ experimentsf36g thenrevealed that currents from the main airstream pass freely inolfactory region during expiration as well. Becker and Kingf210glikewise found different pathways for inspired and expiredNeuhausf211–213g, however, was the first to carry out extensresearch on canine olfactory acuity in the 1950s. He found tdog can detect 1 mg of butyric acid dispersed throughout 108 m3

of air, i.e., the volume of an entire town. Negusf194,206g made abroad comparison of olfaction among many species with theas chief sniffer. Syrotuckf125g examined canine scenting atracking in terms of physical and chemical phenomena. Hemated that the detectable “ground scent” left by a human mlast 8–16 h. Zuschneidf61g discovered the long canine snifNeuhausf89g further examined the anatomy of the nasal passand the role of sniffing in olfaction. Schreider and Raabef64gshowed 29 transverse sections through a beagle’s nose, revthe elaborate scrollwork of the turbinates. Recently Johnstonf214,215g and Williams et al.f216,217g developed laboratory o
factometry for dogs and measured the dog’s response to tra
198 / Vol. 127, MARCH 2005

u-o,-e

ill

n-

f-

ec-e

thec-nd

s.r toll

a-

len

rne

rd.stgn-l-

,irn

he

.

t a

g

ti-ht

es

lingal.

chemicalsssee also Sec. 6.1d. Thesen et al.f218g and Steen et af219g studied canine olfaction in outdoor tracking experimeMorrison f62g and Johnston et al.f103g reported basic expements on the canine olfactory system, including tomographthe nasal cavity and a particle deposition study. Most receSettles et al.f47g visualized the airflows associated with thenine external nares during sniffing. Based on this historical remodern anatomical references now reveal the detailed instructure of the dog’s nosef220,221g.

The canine external nares are shown in Figs. 11sad and 11sbd.Negusf194g first noted that mammalian nostrils give directionthe inspired air, then Stoddartf3g found that the nasal swell bosalar foldd just inside the nostril controls the direction of theflow into the nose. The high-speed video observations underFigs. 11sad reveal the opening of an “upper orifice” above thefold during the inspiration phase of sniffing. Upon expirathowever, this pathway closes, the nostril flaresfFig. 11sbdg, andair is ejected ventrally and laterally through the midlateralsnasal sulcaed f47g.

Figure 11scd f62g shows the upper and lower airwaysdorsalandventral meatusd leading from the external naris into the maxiturbinate region of the canine nasal cavity. The external naristion just described, driven by several olfactomotor musclesf220g,causes the dorsal meatus to receive inspired air and the vmeatus to deliver the spent air for expiration. The dorsal anddal direction of the inspired air channels it directly towardolfactory region of the dog’s nose.

The expired air jets, on the other hand, are vectored bshape of the “nozzle” formed by the alar fold and the flared nowings snasal alad, Fig. 11sbd. Thus the external naris acts avariable-geometry flow diverterf47g. This has three advantags1d it avoids distributing the scent source by expiring back towit; s2d it stirs up particlesfFig. 1sbdg that may be subsequeninspired and sensed as part of the olfactory process; ands3d itentrains the surrounding air into the vectored expired jetsfFigs.1sbd, 1scd, and 12g, thus creating an air current toward the nfrom points rostral to it. This “ejector effect,” shown in Fig. 12an aid to olfaction: “jet-assisted olfaction,” in other words.

We also observed that the “reach” of the canine nares dsniffing is up to about 10 cm, though the dog always narrowsdistance essentially to zero if allowed. Since the nostril inflo

Fig. 11 External nares of a Golden Retriever „a… during inhaleand „b… during exhale portions of sniffing cycle †47‡, and „c…solid cast of the nasal cavity of a dog reconstructed from CATscans †62‡ „cast provided courtesy T. S. Denny Jr., AuburnUniversity …

ceomnidirectionalsSec. 2.4d, the detailed spatial distribution of a


t inrryThtoatno

icattaithss

ow. Wontim

os

ts toitionhav-es apar-the

g asnevitythever,

l cav-rtantr and

rticle

t toicalpres-n 20

h

op-leres.st norly actory

el nos-atorystrilsomi-den-ualt theys and

veraln.

annot

is ama

late

scent source is only discernable when the nose is broughvery close proximity with itf47g. Caninesf218,222g and othemacrosmatic animalsf223,224g need to “read” detailed olfacto“messages” such as scent overmarks left by other animals.ability to do this requiresproximity sniffing, and is analogousour visual reading of text. So, in order to properly interrogchemical traces it really is necessary for a dog to poke itsinto everyone’s business.

Proximity, however, is not required in air scenting of chemtraces borne by plumes, another canine specialtyf125g. Steen eal. f219g discovered that an air-scenting bird dog can maincontinuous nasal inspiration for up to 40 s. They proposeopen-mouth respiration produces a low nasopharyngeal preto induce this continuous olfactory airflow.

Negusf194g suggested that macrosmatic mammals sight dtheir long snouts to focus upon food they are about to seizeobserved, however, that a dog approaching a scent sourceground first scanned its vicinityf47g. Instead of aiming directly athe source, the nose was instead lowered to close nostril proxwith the ground before reaching the sourcefFig. 12sadg. Then thedog advanced toward the scent source, pausing when the n

Fig. 12 „a… Schlieren image revealing the “ejector effect” ofexpired air jets during canine sniffing that draws in air from awarm forward scent source and „b… diagram illustrating the re-gion of expired jet impact upon a ground plane „dotted line …

and the induced airflow caused by jet entrainment †47‡. Seealso Fig. 1 „b…

were directly overhead, sniffing all the while. Often the dog


to

eir

ese

l

nature

nethe

ity

trils

scanned past the scent source, allowing the expired air jeimpinge directly upon it. Finally the nose returned to a posdirectly above the source for a few more sniff cycles. This beior promotes visual as well as olfactory inspection and providlocal survey of the spatial scent distribution. It also disturbsticles in the vicinity of a scent source by the impingement ofexpired air jets.

Syrotuck sees this environmental disturbance during sniffinan aid to olfactionf125g. Morrison examined the uptake of fis0.5 mm–5 mmd charcoal powder inside the dog’s nasal cafollowing sniffing f62g. Particles were found predominantly inanterior nasal cavity ventral to the maxilloturbinates. Howeupon vigorous sniffing, particles reached as far into the nasaity as the olfactory ethmoturbinates. Moist mucosa are impohere, since moisture is the solvent that carries both vapoparticle-borne chemical traces to the olfactory receptorsf125g.When the dog’s nose is wet and cold, it can act as both a paand a vapor trap.

Finally, the warmth of the expired canine air jets may acvolatilize latent chemical traces on surfaces. Many biologodors vaporize at or near body temperature, and the vaporsure of TNT, for example, increases by a factor of 4 betweeand 30°Cf225g.

4.3.2 Cat. The cat’s nose, Fig. 13sad, is in stark contrast witthat of its canine relatives. Though macrosmatic, the entireFel-idaefamily socelot, domestic cat, panther, puma, lynx, Asian leard cat, caracal, and bay catf226gd appears to lack the variabgeometry and multifunctionality of the canine external naWhile a proper aerodynamic study has not been done, almonostril motion is observed when the domestic cat sniffs. Cleasophisticated canine-type nostril is not a prerequisite for olfaability, even among carnivores.

4.3.3 Horse. The horseEquus caballusand its relatives arodd-toed ungulates and prey animals. They have wide ovatrils and long straight nasal passages allowing a high respirairflow rate while running, when dilator muscles cause the noto flare. The nostrils contract upon expiration to produce prnent ventrally-expired jets that are visible due to moisture consation in winter. Horses sniff to identify food, kin, and sexstatus, and to acquire olfactory warnings of the predators thasuspect are lurking behind every bush. Unlike dogs, horsetheir relatives have dry rhinaria.

4.3.4 Pig. The swine nose is spade shaped and has seuses. Moultonf207g considers it to be a “chemotactile” orgaLike elephants, pigs use their snouts for tasks that their feet caccomplish. The pig nostril inlet orifice, Fig. 14sad, is wellrounded according to the discussion of Sec. 2.4. In fact, itsimple flanged bellmouth inlet with a hint of an inverted-comshape.

4.3.5 Cow. Like pigs, cows are members of the great ungu

Fig. 13 External nares of „a… a domestic shorthaired cat and„b… a burro, Equus asinus , photo by L. J. Dodson

prey orderArtiodactyla, which also includes the antelope, deer,

MARCH 2005, Vol. 127 / 199

mrte4s ton-

olatughmestr

udeus,or

h, nin

ty o

el-qum

nke fp--liknissturl-iffe

tctlymti

Aatioarize

e rsobitri.

bubo

nt

the-

nd itswardof a

ctedal-

solu-d oneint atthese oftent,ted

resses

eowricateion.ody-lencetheletsuch

lts are

anser theeents, sotheg itinucing

sci-ow a. Lab20%

, 1/2onalb

matic,atory-

goat, sheep, camel, caribou, moose, giraffe, and hippopotaMany of these animals have opposed, laterally oriented, invecomma-shaped nostrils as seen distinctly in the cow, Fig. 1sbd.The inverted-comma nostril shape, upon constriction, narrowair passage to a crescentf194g. This suggests some distant cnection with the canine midlateral nostril slit, but in fact nonethe prey animals has a nostril of sufficient mobility to moduthe airflow in a canine fashion. At least two ungulates, thohave variable-geometry nostrils for other purposes: The canostrils close to keep out the sand, while the moose’s noclose underwaterf227g.

4.3.6 Bat. The bat’s naris is some two orders of magnitsmaller than that of theBovidaejust discussed. The insectivoroLittle Brown Bat, Myotis lucifugus, common in North Americahas broad nostrils with an inverted-comma shape. The elabnose “leaves” of some bat species are for ultrasonics, thougolfaction. The bat’s forwardly oriented nostrils aid olfactionflight by ram-air sampling. Some bats can discern the edibiliinsects they pursue by using “wake olfaction”f3g.

4.3.7 Elephant. The elephant’s trunk is a unique and cebrated multipurpose chemotactile organ. Elephants aremacrosmatic even though their external nares and olfactorycosa are separated by a considerable length of trunk. The truonly explores proximity scents but also acts as a periscopdirectional air scentingf228g. Other trunk functions include grasing food, drinking, spraying water, respiration, greeting, handtactile operations, trumpeting, and fightingf228g. Its cross sectioshows two oblong central nasal tubes formed of connective tand lined with mucosa. These open into the elephant’s nasonates at their posterior extremityf229g. As noted earlier, the eephant is a natural prototype for hand-held wand-type snprobes.

4.3.8 Rabbit. Glebovskii and Marevskayaf230g found tharabbit nostril motion, driven by the narial muscles, is direconnected with high olfactory brain activity. Inspiration is accopanied by nostril dilation, reducing airway resistance. “Theandalae nasimove upward while the floor of the nostril sinks.”sharp rise in airway resistance then signals the onset of expirGlebovskii and Marevskaya further describe the rabbit nmuscles as “the propriomotor apparatus of the olfactory analyZwaardemaker mirror tests by Bojsen-Møller and Fahrenkrugf31gshowed that expired air is directed ventrally and laterally in thand rabbit, as in the dog. The rabbit’s nose, Fig. 15sad, resemblea cap over a vectored duct. Little else is known about the mnostril function of this macrosmatic mammal. A rabbit nosaerodynamics study along the lines off47g is certainly called for

4.3.9 Human. Our nose is retrograde and microsmatic,heavily investigated. Sources on human nasal airflow includepopularf231g and scholarlyf232g accounts, historically significaworks f2,233g, and recent review articlesf5,234–236g. The de-

Fig. 14 External nares of „a… a piglet and „b… a calf, photos byL. J. Dodson

cline of the primate olfactory genes leading up to humankind

200 / Vol. 127, MARCH 2005

us.d-

he

fe,

l’sils

ateot

f

iteu-notor

e

uebi-

r

-p

n.alr.”

at

lel

tth

also well documentedf237,238g. Though less elaborate thanother animal noses, our nosefFig. 15sbdg is still sensitive, interesting, and comparatively well studied.

Section 3.2.1 described the rising human boundary layer aparticles. We stand erect and our nostrils are directed downso they inevitably sample this boundary layer, unlike the casefour-legged animal with a long snout. Expired air jets are direventra–rostrally and are usually fast and turbulent, allowingmost no rebreathing of the exhaled airf239g.

Keyhani et al. carried out a steady laminar Navier-Stokestion of airflow into the human nasal cavity with Re=610 basethe external-naris hydraulic diameterf240g. Results show that thinternal path taken by inspired air depends upon its entry pothe nostril, with only the anterior tip of the orifice supplyingair that eventually reaches the olfactory epithelium. Becauthis, 90% of the inspired air is not sampled for olfactory conas already known from work cited earlier. The flowfield compuby Keyhani et al. was then used to derive an olfaction modelf240gthat yields the odorant mass flux sensed by the nose and addsome of the questions raised in Sec. 2.3.

Subramaniam et al.f16g did a similar CFD solution at R=1360 for an inhalation toxicology study. Here, recirculating flin the nasal vestibule and nasopharynx produced a more-intoverall flow pattern with even less airflow to the olfactory reg

In vitro flow visualization experiments on human nasal aernamics were described in Sec. 2.1. The appearance of turbuin some of the resultsf241g, however, raises questions aboutlaminar-flow CFD simulations just described. A low nostril inRe does not guarantee laminar flow throughout, especially incontorted passages, but no turbulent or unsteady CFD resucurrently available.

Finally, Sobel et al.f242g study the neurophysiology of humolfaction in vivo, finding that it is not the movement of nomuscles that primes the human brain for olfaction, but ratherush of air up the nose. Differences in the airflow rate betwnostrils cause each nostril to be sensitized to different odoraneach nostril conveys slightly different olfactory information tobrain f242g. Further, sniffing for a scent and actually smellinactivate different brain regionsf243g, and sniffs are modulatedresponse to odor content, higher odorant concentrations indlower-volume sniffsf63g.

4.3.10 Rat. The laboratory rat holds a unique position inence. For reasons similar to those given for humans, we knlot about rats although, unlike humans, rats are macrosmaticrats have perhaps 1500 olfactory genes of which only someare “junk” DNA f237g.

Human carcinogen response is assessed by two-yearmillion-dollars-each rat studies carried out by the U.S. NatiToxicology Programf112g. Extrapolation of the results from larats to humans to assess human health risks is problethough, given large interspecies differences in nasal respirphysiology and airway anatomyf16g. Inhalation toxicology nev

Fig. 15 External nares of „a… cottontail rabbit, Sylvilagus ob-scurus and „b… human, Homo sapiens

isertheless uses live rats or their nasal molds to measure the depo-


stox

owth

ofa tsmendes

etitcharer--an

ardageat’snlyf

o-

m-sannta

ltip-

bleltoon

therares, fishood,f fish

epa-ior

withc.useiresridge

over-taticy bejaw

ions.te-

nde

eratesharkof thesen-

er op-ark

ngingfing,catede ol-gisorne

pos-

is ispro-

ese

ficesoilThe

sition of respirable particulatesf244–246g. The albino lab rat ithus the key to determining safe levels of human exposure toinhalants.

A CFD solution of the Navier-Stokes equations for the airflin the passages of a lab rat’s nose yields some insight intoprocess, Fig. 16sad andf53,247g. As in dogs and humans, mostthe inspired air bypasses the rat’s olfactory zone and exits vinasopharynx. However, the airstream that actually reachesensory ethmoturbinates enters dorsally, reverses directionally, and finally exits ventrally through the nasopharynx as icated by arrows in Fig. 16sad. Of course CFD only approximatthe convoluted geometry of an actual rat’s nose, Fig. 16sbd f248g.

Like the rabbit and dog, the rat has active, variable-geomexternal nares. In both ground and air scenting the nose pand the nostril twitchesf249g. In some rats the external nares fland the nose hairssvibrissaed move rhythmically, projecting duing inspiration and retracting during expirationf3g. The inversecomma-shaped narial orifice directs expired airflow ventrallylaterally, causing relatively little anterior air exchangef250g andforming two fan-shaped areas of condensation on the Zwamaker mirror, Fig. 1sad f31g. A bout of sniffing grows to yieldmaximum airflow rate at its conclusion. Youngentob et al. suga “sniffing index” describing many characteristics of the rsniffing strategyf251g. During a discrimination task, rats suddeswitch from a low s2–5 Hzd sniffing rate to a high rate o8 Hz–10 Hz for unknown reasonsf252g. Here, too, a proper aerdynamic study of external nostril airflows is needed.

4.3.11 Opossum. While dogs and cats diverged from a comon ancestor some 50 million years ago, the opossum hachanged significantly in 90 million years. He is both ancientmodern, since he continues to compete effectively with placetoday. His chief current challenge is to stay out of the road.

The opossum has the long, well-equipped snout and muturbinates of a macrosmatic animalf194g. The external naris aperture, Fig. 17sbd, is laterally oriented and not obviously variain geometry, although there are no known studies of it. Mouf207g notes that olfactory structures are most prominent am

Fig. 16 „a… CFD solution of airspeed in the F344 rat’s nasalcavity †53,247‡, courtesy J. S. Kimbell and „b… sectionalanatomy of an albino rat’s nasal cavity with darkened olfactoryepithelium †248‡, courtesy J. S. Kauer

older animals like the opossum, not in the higher primates an


ic

is

hethedi-

i-

ryes

d

e-

st

notdls

le

ng

aquatic mammals.

4.3.12 Fish. Here we leave the mammalia to consider overtebrates, many of whom have simple immobile external ncompared to those of the dog, rabbit, and rat. Like mammalsneed their sense of smell for habitat and kin recognition, freproduction, and predation avoidance. Detailed accounts oolfaction are given by Negusf194g, Kleerekoper f253g, andStoddartf3g.

The key issue in fish olfaction for present purposes is its sration from the respiratory gill flowfield. Anterior and posternaris holes or slits supply through flow to a chamber filledlamellae that provide a large sensory surface areassee also Se5.3.1d. To overcome the internal olfactory pressure drop, fisheither passive or active water circulation. The former requrelative motion between fish and water, where sometimes aor scoop just aft of the anterior naris assists olfaction by recing dynamic pressure there while the flow vents to local spressure downstream. Active olfactory water circulation madriven by cilia or may occur as an accessory to respiratorymotion.

Some fish feature highly specialized naris adaptatPolypterus bichir, for example, has a fluted Pitot-tube-like anrior naris extensionf3g, while in moray eels both anterior aposterior nares may be extended. SomeTetraodonspecies havtheir entire olfactory apparatus out on a stalkf253g.

Shark olfaction is almost as adept as their ability to geninterest and fear in people. The Great White and many otherspecies have flush anterior nares located on the undersidesnout, feeding water to an olfactory chamber of spectacularsitivity. Sharks track prey wakesf254g by the chemical plumtracing methods described in Sec. 3.3.3. Here is yet anotheportunity to study, for the first time, the fluid dynamics of sholfaction.

4.3.13 Birds. Birds sometimes have only rudimentary sniffiapparatusf3g, but in other species olfaction is key to their feedand nesting behavior, and even to navigation. Airborne sniffor example, allows scavengers like the turkey vulture to lotheir malodorous meals. Atmospheric trace gases also provifactory navigational cues to some birdsf255g, the young learninprevalent trace gradient patterns at homef256g and then using thknowledge to navigate by olfactory detection of natural airbvolatiles f257g.

Many birds have simple round or oval nostril holes in theterior region of the beak. They also have turbinatesf209g thatfunction much like those of mammals. Sometimes the narcovered by skin or feathers, and often the anterior turbinatetrudes into the orificef195g. Such obstructions suggest that thspecies are not keen sniffers.

Only one bird, the highly specialized kiwi, has its nostril oriat the tip of a long beak like a wand-type sniffer. It probes thefor worms, so a posterior naris location would be worthless.

Fig. 17 „a… External nares of a Sprague-Dawley lab rat pro-vided by M. J. Kennett, Penn State University and „b… externalnares of the opossum Didelphis virginiana

dbeak tip, Fig. 18sad, consists of an anterior probe, a longitudinal

MARCH 2005, Vol. 127 / 201

inar

reltemofcudy

rea, also

pegu

crerde

n-lythan

rprdileanr o

edrnad

-beounand

ion

few

c.ica

gelcer t

minesis Reown-

when

thet-peed,lity toamic

s in

ynoldsnnaenot

se ofnalsnaeady

ter-

toesined

anda

ts

n aan

rac--palpso

pic., for

bean

slit nostril, and a lower mandible. One wonders if naris pluggby soil and olfactory signal deposition in the long nostril tubesworrisome to the macrosmatic kiwi.

TheProcellariiformesf195,258g are seabirds that include petand albatross species, all with highly developed olfactory sysand distinct tubular nostrils. They can follow the trace odorfood source for many km over the ocean. Following the dission of airborne sampling in Sec. 3.3.1, it is hard for a fluidnamicist to see such nostrils, Fig. 18sbd, without thinking “Pitottubes.” These nostril protrusions recover the impact pressuvelocity head 1/2rU2 during flight, automatically providingpressure boost to overcome losses in the olfactory systemdoing so without additional effort from the bird. Pennycuick asuggests a related Pitot-tube function—that of an airsindicator—allowing the bird to sense and take advantage ofenergy in the separated flows downstream of oceanic wavef259g. Procellariiformes tap this unlimited energy source in oto soar almost effortlessly over long distances.

4.3.14 Reptiles and amphibians. Reptiles and amphibians geerally have rudimentary olfactory apparatus. Most have onsimple pair of external naris holes flush with the front or top ofhead f3g. Olfactory power varies among species, though,some lizards, snakes and crocodilia are macrosmatic. Watenaris closure and “snorkel” nostrils adorn the newt, crocosalamander, and frogf3,260g. Frogs even have flap-separatedterior and posterior olfactory regions suited for sniffing in wateair, respectivelyf237g.

Not much is known about dinosaur olfaction, but Witmerf261gplaces theT. rexexternal naris well forward of its usually-depictlocation in order to provide room for a more significant inteolfactory apparatus. AlsoRhomaleosaurusis believed to have haNational Advisory Committee for AeronauticssNACAd-scoop-type naris inlets inside its mouth, venting externallyf262g.

4.4 External-Flow Noses of the Invertebrates.Here the sensor chamber of the vertebrates is turned inside out: no chamall, just sensor-bearing antenna stalks extending into the surring fluid. Treelike sensor structures have evolved in both airwater to present a large surface area for olfaction. The fluidnamics of theseaesthetascshas received a lot of recent attentf93,170,176,196,263g.

For brevity we consider here the morphology of only ainvertebrate sniffers from the phylumArthropoda.

4.4.1 Crustaceans. Lobster “sniffing” was introduced in Se2.7 in terms of diffusion and in Sec. 3.3.3 in terms of chemplume tracing. Lobsters have twin antennae with lateral flabearing fine sensory aesthetascs, Fig. 19. There are differenspeciesf41g, but in general the antennae are flicked in orde

Fig. 18 „a… Close-up image of the Great Spotted Kiwi’s beaktip, copyright Chris Smuts-Kennedy, reproduced by permis-sion. „b… The upper beak of the tube-nosed Dove Prion, Pachyp-tyla desolata , redrawn from †258‡

sample the environmentf92,93g. This sniffing behavior is gov-

202 / Vol. 127, MARCH 2005

ge

sas--

or

nd

edststsr

aedoof,-r

l

r atd-d

y-

llas ino

erned by the aesthetasc Reynolds number Re which deter“leakiness,” i.e., how much water passes between them. Thtends to be near unity, so the lobster can sample with a fast dstroke and hold the signal with a slower upstrokef176,264g. Pé-clet numbers are large for these motions, but diffusion actsthe antennae are still.

Similar principles apply to other crustaceans, includingcrayfish, crab, and Mantis shrimpf41,65,177,196g. These soughafter species take advantage of zones of high mean flow swhich increases turbulence and degrades a predator’s abitrack a chemical plume. Thus fast flows provide a hydrodynrefuge from olfactory-mediated predationf179g.

4.4.2 Insects. Chemical plumes are as important to insectthe air as to crustaceans underwaterf168,265,266g. Moths tracinga pheromone plume are the classic example. The same Renumber effect on the apparent porosity of branched antef90,267,268g arises, though the comblike moth antennae doflick and are spectacularly different in appearance than thothe lobstersFig. 20d. Some moths respond to pheromone sigby flapping their wings to induce airflow through their antenwithout flying f268g, an aspect of insect behavior that has alreattracted biomimicryf182g.

Here briefly consider mankind’s worst enemy, that supremerorist and chief vector of biological warfare, the mosquitosFig.21d. With a human death toll of over one million/year, mosquias killers easily outrank all the worst human tyrants combf269g.

Mosquito host location is known to be an insect-sniffingchemical-plume-tracing issuef168,270,271g. Humans give offvariety of kairomones, chemical signalssCO2, lactic acid, etc.dthat attract mosquitoesf272g. The thermal convection currenproduced by vertebrates carry these trace chemicalsf270,273g,and mosquitoes fly upwind to locate their sourcesf180g. Anoph-eles gambiae, for example, flew an upstream zigzag path iwind tunnel odor plumef272g. In another study, smelly humfeet were a noteworthy mosquito attractantf274g.

An important human goal is to interrupt mosquito-host intetions, thus inhibiting the spread of diseasef275g. The hostlocation sensilla are on the mosquito antennae, Fig. 21, andf272,276,277g. At normal flight speed a 1-mm-diameter mosquitsensillum hair has a Reynolds number of about 0.06.

An outsider needs extra caution around this intricate toThere are many mosquito species with different behaviorsexample f272g. Nonetheless some fluid dynamics seems tomissing here, namely:s1d a proper understanding of the hum

Fig. 19 Lateral flagellum of one antennule of the clawed lob-ster Homarus americanus . Proximal diameter shown is 1.4 mm.Hairlike projections are aesthetascs and guard hairs „out ofwater and in disarray …. See †92‡ for electron microscopy images

Fig. 20 „a… Giant silkworm moth, Antheraea polyphemus ,
wingspan 8 cm and „b… closeup of feathery antennae

Fosce

ms-, afitth

peyer

a

alwa

hhactio

e,ar

ecthatr

uce

emo-sesoseide at byares

ratestion

n oflentks inbout

sby ae ofgase con-werement

theyaving

r-n of

inivesorsces,

ox-, and

ench-eredrthernly-

n beMi-but noec-

d or

ge. Aff airs-sam-

d-ensor

inellessd itsut,even

, andIMSvail-etec-

thermal plume and wake behaviorsSecs. 3.2.1 and 6.3.3d and s2da knowledge of the mosquito’s flight envelope and limitations.example, in one popular book on mosquitoes the humanplume is said to be heavier than the surrounding airf269g. Normalflight speed is quoted at 1 m/s and maximum speed at 1 1/3f269,278g, yet there is little information on the inhibition of moquito bites as a function wind speed or host motion. Finallythough our knowledge of insect aerodynamics has benegreatly from studies of the innocuous dragonfly, where areparallel studies of mosquito aerodynamics? The mosquito apto be a fragile insect, all rickety and strung out like a Wright FlIf we knew more about her aerodynamics, it might suggest wto push her beyond her flight envelope.

4.5 Directional Olfaction. Von Békésy claimed directionodor perception if as little as a 10% concentration differencepresented to the left and right human nostrilsf279g. Later researcquestioned thisf280g, but there is still considerable evidence tat least some animals practice and depend upon “stereolfaf3,242,281g. In addition tochemotaxisandanemotaxis, describedearlier, the termsklinotaxis, swinging the head from side to sidandtropotaxis, stereolfaction via widely-separated naris inlets,used.

Stoddart f3g suggests that nostril adaptation goes with nstiffness, and that stiff-necked fish need tropotaxis moresupple-necked land species. Tube-nosed bats appear to usetactic olfaction in flight, where turning the head would prod

Fig. 21 „a… Mosquito feeding on a human hand. CourtesyPhilip Myers, http://animaldiversity.ummz.umich.edu. One an-tennal flagellum is visible. „b… Microgram of a mosquito anten-nal flagellum, about 20 mm in diameter, showing two completesegments, the long sensilla chaetica , and the numerousshorter sensilla trichodea , photo by J. M. Listak

unwanted aerodynamic forces. Dogs practice klinotaxis, movin


rnt

/s

l-ede

ars.ys

s

tn”

e

kn

opo-

their heads while tracking or scenting, and probably also antaxis based on the differential wind cooling of their wet nof125,218g. Moulton f207g, on the other hand, notes that the clnarial spacing of many mammals does not apparently provsufficient lateral signal difference to discern an odor gradientropotaxis. The myth of tropotaxis due to the widely spaced nof hammerhead sharks was debunked by Kajiura et al.f282g, whofound that an anterior groove on the shark’s cephalofoil integthe olfactory input, thus providing no more tropotactic separathan that of any other shark species.

Finally, Atemaf42g demonstrated that the lateral separatiolobster antennae is sufficient for “eddy chemotaxis” in turbuodor plumes. He also found that the sharpness of odor peapassing eddies can provide “odor landscape” information athe lobster’s location relative to the plume’s source.

5 Artificial Olfaction

5.1 Artificial (Electronic) Noses.Artificial noses are devicefeaturing several nonspecific odorant sensors interrogatedpattern-recognition system in order to identify a broad rangodorantsf9g. This definition excludes specific sensors likeleak detectors and mass spectrometers, but these will also bsidered later. Still very much a work-in-progress since theyintroduced a few decades ago, artificial noses already auglimited human olfactory skills by being less subjective, andmay even supplant the trained canine sniffer someday by hsimilar sensitivity but being willing to work longer hours.

5.1.1 Overview of current artificial noses. Gardner and Batlett f9g give a brief history of electronic noses and a discussiocurrent commercial devices. Similar overviews are foundf283,284g, while Yinon f285g summarizes “e-noses” for explosdetection. Artificial noses currently have nonspecific senbased on a variety of principles including quartz microbalansurface acoustic wavesSAWd technology, capacitance, metalide semiconductors, calorimetric or amperometric sensorsphotopolymers.

Most of the present commercial devices are ponderous btop analytical instruments that do not sniff and are not considfurther here. A few portable or handheld noses deserve fuattention, but most of them also lack any real ability to sniff. Othe Smiths Detection Cyranosef286g has both an internal sampling pump and a wand-type inlet with a flared end that caconsidered a sniffer. The Applied Sensor VOCcheck andcrosensor Systems Vaporlab handheld e-noses have pumpssniffers sapparently only fittings are provided for direct conntion to an odor sourced.

The Nomadics “FIDO” artificial nose can be either handhelwand-mountedsfor landmine detectiond f287g. Ambient air isdrawn through the sensor, after which there is a clean-air pursimilar prototype device employs a small vacuum tank to sniat ground levelf288g. The Z-nosef289g uses SAW and gachromatograph technology in a hand-held sensor with a 10 spling time. Gelperin’s e-nosef290,291g inhales the air surrouning an object through a perforated platform located over a sarray.

The Tufts Medical School/CogniScent Inc. Scentf6,292,293g is not yet a commercial device, but is nonethenotable for the biological inspiration behind its sensors anpattern matchingf294g. Moreover it is a true sniffer with a snoa flanged inlet and a through-flow sensor chamber. It caninhale and exhale like an animal.

5.1.2 Related nonartificial-nose handheld detectors. A popularand useful nonartificial-nose detector for explosives, drugschemical agents is the ion mobility spectrometer orf295,296g. At least two commercial handheld devices are aable, the GE Security VaporTracer, Fig. 22, and the Smiths D
gtion Sabre 2000. Both have internal pumps to acquire air samples,
MARCH 2005, Vol. 127 / 203

omum

WstebuIncth

croow

es

siv

ers

the

a-

h

po

a cth

-mic

thed

o-ticlan

vesrpuxpl.sus.ticlbof thinduct

m

uffersde-y be

au-lowto anrther,

re of

ctionnglyar-

ncede-

ves a

istryfilledel or

to re-ec.ighand

eation

r airin a

ehanceatedrrier

reamciallyation

orm,rs.a areachedon-

of the

.ist-t andodor-ration

inceniffered byitduceEx-

sinceven

of aN.

aux-fig--

but the volume of air that an IMS can directly accept is only sml/min. A third IMS device, the Implant Sciences QuantSniffer, takes a different approach discussed in Sec. 5.2.4.

The Sandia Microhoundf297g uses a micro-IMS and a SAsensor, and inhales air through a flanged inlet. The MesosyBiocapturef298g impacts and liquefies samples for analysis,details of its sniffer are not available. Finally, two Thermedicspatentsf299,300g describe handheld shrouded sampling gunsapply heat and air-jet puffers to surfaces, then inhale air aa preconcentrator. These devices are without any kncommercialization.

5.1.3 Desired characteristics of sniffers for artificial nos.Ideally an optimum sniffer ought to:

1. Have sufficient “reach” to acquire the desired explotrace signal without physical contact,

2. Localize the scent sourcesexcept for those special snifftypes designed to sample broad areas like the sidevehiclesd,

3. Display immunity to ambient conditions, especiallybreeze,

4. Have simplicity, mobility, light weight, small size, and resonable power requirements, and

5. Have the ability to disturb surface environments enougdislodge and collect particles as well as vapor traces.

5.2 Airborne Trace Sampling for Artificial Olfaction.

5.2.1 Headspace sampling versus sniffing. In most benchtoinstruments of analytical chemistry, including the majoritycommercial e-noses, the saturated vapor above a sample intainer is drawn by carrier gas through tubing into the input ofinstrument. This is headspace samplingf301g, and while it is essential in laboratory practice, it is distinct from aerodynasampling—sniffing—as defined here. Few animals haveluxury to squat on a lab bench, have odor samples deliverthem, and suck the samples through a straw.

5.2.2 Particles versus vapor. There is a controversy in explsive detection circles over whether one actually detects paror vapor traces or both. A key paper on this topic is DavidsonStott f302g, who draw a distinction between volatile explosisvapor pressure higher than nitroglycerind and nonvolatiles: vapodetection of nonvolatiles usually depends upon solvents, imties, or breakdown products, whereas the vapor of volatile esives like tri-acetone-tri-peroxidesTATPd is directly detectableSuch volatile vapors are readily adsorbed, however, bythings as the packing materials found inside cargo container

In general we should be ready to sniff and detect both parand vapors, since they provide complimentary information athe odorant source. Particle sampling is the more difficult otwo, but it has the advantages of larger signal levels and ofcating the actual chemical species, not just its vapor byprod

Approaches to resuspend particles from surfaces include

Fig. 22 The GE Infrastructure Security VaporTracer sniffing abriefcase

chanical agitation and vacuuming, swabbing, vibration, therm

204 / Vol. 127, MARCH 2005

e

mst.atssn

e

of

to

fon-e

eto

esd

ri-o-

ch

esutei-s.e-

desorption, shock waves, pulse-pressurization, and jet psSec. 2.5d. Wrapping and handling an improvised explosivevice, for example, is likely to generate a particle field that madetectable even if the vapor of the explosive is notf303,304g.

Griffy’s calculationf94g was mentioned in Sec. 2.7 with a ction about giving up on vapor detection. Vapor diffusion is a sprocess, but some sniffers can get in such close proximityodorant source that they can get a good sniff even so. Fuvapors are readily transported in thermal plumesf48g.

On the other hand, given the extremely low vapor pressuexplosives like pentaerythritol-tetranitratesPETNd and RDX, it iscertain that particles, however small, play a role in the deteof these substances by sniffing. Nonvolatile explosives stroadsorb to solidsf305g, including the ever-present airborne pticles in the human thermal plumesSec. 3.2.1d.

5.2.3 Preconcentration. Preconcentration and the “impedamismatch” were introduced in Sec. 2.3, while impactors werescribed in Sec. 3.1. How it is done in practice, though, deserlittle further elaboration.

Traditional preconcentration methods of analytical cheminvolve passing an air sample through a permeation tubewith a sorbent material, such as activated charcoal, silica gTenax™. That material is subsequently thermally desorbedlease trapped chemical tracesf306–309g. Personal samplers, S3.2.2, often work this way. But given the small size and hpressure drop of a typical sorbent tube, this approach is slowineffective for high-flow-rate sniffing.

Membrane filters can serve the same purpose as permtubes and can be desorbed in a similar fashionf13,101,310g, butare likewise not suited for high flow rates.

A third approach, more adaptable to rapid sniffing of largevolumes, is to collect the trace odorant on a metal surfaceflow-through preconcentratorf310–313g. Depending upon thodorant, various treatments can be applied to the metal to enmolecular or particle deposition. The metal can then be heelectrically to quickly desorb captured traces. A secondary cagas flow is needed during desorption, when the main airsteither stops or is shunted away. Cold metal surfaces espeattract trace chemicals like explosives, but moisture condensis an associated problemf314g.

When the desired airborne trace material is in particulate fimpactors or cyclonessSec. 3.1d make suitable preconcentratoFor example, biological pathogens like spores and bacteriusually impactable, as are explosives or chemical agents attto human skin flakes or textile fibers. Virtual impactors preccentrate such particle-laden airstreams by discarding mostairflow while keeping the particles.

5.2.4 Jet-assisted olfaction. One of nature’s lessonssSec4.3.1 and Figs. 1 and 12d is to put exhaled air jets to work assing the sniffing process. Thus the dog and probably the rarabbit can disturb surface particles, inhale them, and desorbants from them in the nasal mucosa—a natural preconcentsystem.

How can we use this knowledge in a man-made sniffer? Srespiration is unnecessary, auxiliary air jets adjacent to the sinlet are required to disturb surface particles, as first suggestMcGown, Bromberg, and Noblef299g. Beyond that, though,was also shown in Sec. 4.3.1 that the dog’s exhaled jets proan ejector effect that can improve the reach of the sniffer.tended reach is just as important for man-made sniffers,proximity sniffing is not always possible and is occasionally edangerous.

At least two schemes are available to extend the reachsniffer, the first borrowed from the field of ventilation. C. P.Aabergf315g invented an elegant inlet reach extender usingiliary jets, Fig. 23. The central inlet, facing downward in theure, takes in flow rateQi while an additional airflow 2Qj at above
alambient pressure discharges laterally forming turbulent jets. The

r intwe

thfo-

locig

ichterecetrava

erpr

kinedn t

owesiotiorabaceThiff

Na

o dy iun

Ifnd

madentact

eas.matethanffec-

, but

rub-d, theessure

low-50%

e. Itrposely to

f an

go thee el-to allere toinum,

n par-hewn-

nlyntrte-d thevitya-llows

-oradialg asdirect-longand

sory

jet entrainment makes this device appear to be a much largewith an extended reach, but only the central streamtube bethe vertical dashed lines is actually taken into the inlet. Thusotherwise-omnidirectional potential-flow inlet is effectivelycused toward the forward direction by jet assistance.

The ventilation community has embraced this concept forfume exhaustsusually axisymmetric rather than planar as in F23d, and several studies are reportedf316–319g. The ratio of thejet to inlet momentum flux is found to control the reach, whcan be several times that without jetsf320g. This jet assist is nofree, but is well worth it in the exhaust of welding fumes, whthe extended reach avoids having the inlet in the welder’s fa

The Aaberg principle has never been applied to chemicalsniffing and is suggested here for the first time. It has the adtage that the large volume of air that must be moved in ordachieve a long reach is not all inhaled, thereby reducing theconcentration chore.

The second jet-assisted sniffer scheme is that of MotchKrasnobaev, and Bunkerf96g. Here the auxiliary jet flow is passthrough tangential nozzles or vanes to produce swirl betweeinlet srightd and the surface being sampledsleftd in Fig. 24. On thecenterline of this “cyclone” a sampling orifice withdraws a flrateQi. According tof96g “The cyclonic motion…creates a tubconsisting of a wall of moving gas that behaves like an extenof the tube that formed the external sampling orifice.” No menis made of a vortex core, but data are shown indicating a favopressure gradient along the centerline from the sampled surfthe sampling orifice over separation distances up to 15 cm.jet-assisted inlet is used in the Implant Sciences Quantum Sna commercial IMS detector that was developed under U.S.funding.

5.2.5 Isolation from ambient wind. All of the sniffing inletsdiscussed thus far, whether jet assisted or not, are subject truption by a lateral breeze. The effect is shown qualitativelFig. 25, where a mild breeze interrupts the sampling of a boary layer by a simple flanged inletf47g.

In the animal world, the only remedy for this is proximity:your nostrils are touching the sampled surface, then the wi

Fig. 23 2D potential-flow simulation of an Aaberg inlet

Fig. 24 Diagram of a cyclone sampling nozzle for an ion mo-
bility spectrometer, from †96‡

letene

al.

.cen-toe-

e,

he

nnletois

er,vy

is-nd-

is

not an issue. However, practical considerations require man-sniffers to have a finite standoff distance and no hard cosespecially in landmine detectiond.

No real solutions to this problem are available, just a few idBaturin,f17g Chap. 6, for example, describes a method to estiinlet streamlines in a crossflow. No remedy is given otherincreasing the inlet suction. Likewise shrouded probes are etive against cross flows in flight and wind tunnels, Sec. 3.3.1are not a cure for the present difficulty.

One approach that might work is a “soft” shroud made ofber, brush fibers, screen, or porous filter material. To succeelateral pressure drop across it must balance the dynamic prof the crossflow. Experiments by Cant, Castro, and Walklatef321greveal that high-porosity screens have little effect, whereasporosity screens act like solid surfaces to the crossflow. Nearporosity, interesting things happen.

An air curtainf322g has also been suggested for this purposmight succeed if its entrainment can be made to serve the puof jet-enhanced olfaction described above; otherwise it is likeexacerbate the problem.

In any case, aerodynamic sniffer isolation from the effects oambient breeze is an important topic for further research.

5.2.6 Signal loss in sampling tubes. Signal loss in samplintubes is an analytical chemistry concern that carries over tpresent topic. Most animals have short sampling tubes, thephant and kiwi excepted. Man-made tubes cannot be inerttrace chemicals, and explosive molecules, for example, adhTeflon, glass, Pyrex, quartz, nickel, stainless steel, gold, platcopper, fused silica, aluminum, and plasticf307,310g. Heatedtransfer lines are used to avoid such deposition except wheticles are being transferredf302g. In that case there is also tconcern that particles will impact and stick to tubing walls dostream of bends, etc.f13,57,323g.

5.3 Sensor Chamber Fluid Dynamics for ArtificialOlfaction. Sensor chamber fluid dynamics is virgin territory: oone previous investigation is knownf324g. Insofar as all preseartificial noses carry their olfactory sensors internally like vebrates, we consider only that case here. Further, we avoiconfusing multifunctionality of the mammalian nasal casolfaction+respiration+air-conditioningd by examining the seprate olfactory chambers of fish. Reynolds number scaling aus to move freely between air and water in this regard.

5.3.1 The eel as a role model. Specifically, consider the common freshwater eel,Anguilla anguilla, as a role model for senschamber fluid dynamicsf3,40,253,325g. As shown in Fig. 26,simple anterior naris on the left leads directly onto the mepassage of a chamber packed full of olfactory lamellae. Actinflow turning vanes as well as sensor surfaces, the lamellaethe water laterally and ventrallysinto the paged. Cilia induce rotation between the lamellae as shown, bringing the flow up athe sides of the chamber to its roof, where it is collecteddischarged out the posterior naris.

This highly three-dimensional flow encounters a large sen

Fig. 25 Schlieren photos of a flanged inlet sampling a thermalboundary layer „a… in still air and „b… with a light lateral breezefrom the right †47‡

surface, ensures that the entire flow is thoroughly sampled, and

MARCH 2005, Vol. 127 / 205

ewt tpr

ne

ellach

thi

rcotiosep, jump

tivensresghrsiffeint

ssyto

s i

aan

s thoidioneelerabur-evmmllywi

geovenng

sa

-

culesnsferalsoower

piri-and

ng

cesx-gy. Inin aThethethat

e flowre instryandodd-

lic di-

ions,arge

, and

ngly

e theare

ading

vesians.ularevas-curitypons

f thef thatemi-ssedunlike

entalning

mag-, but

in-t pur-brief

rnedine,gthatdogstificialast one

does it all in a compact volume.sThe scale shown is at most a fcentimeters, but could be an order of magnitude larger in air asame Re.d This compactness is an asset to the animal and itvides a quick olfactory response time. Artificial nose desigcan scarcely find a better natural example than this.

Though air-breathing vertebrates favor turbinates over lamthis sort of olfactory sensor chamber is nonetheless broadlyacteristic of all vertebrates. According to Stoddartf3g, “There isan inlet to, and an outlet from, a chamber in which is held asensory membrane of sometimes immense area.”

5.3.2 Pressure losses, fans, and pumps. In Sec. 2.3 the flowlosses in a sensor chamber and the energy required to ovethem were introduced. These pressure losses need to be fricas required by the analogy described below, not due to flowration. Separation losses do not arise from effective olfactionbad design. Principles of good design can be found, for exain f326,327g.

In artificial olfactory systems a pump or fan supplies the mopower to overcome losses and induce a flow through a schamberf14g. Such a device produces its maximum overall psure differentialsheadd when there is no flow, thus no loss throuthe sensors. As the flow ratescapacityd Q through the sensorises, the pump or fan’s ability to sustain a given pressure dential drops. Where the two curves meet is the operating pothe olfactory system. Every undergraduate fluids engineeringdent learns to draw such head/capacity curves for pumpingtems. Designers of artificial noses should learn to draw themssee, e.g.,f49gd.

5.3.3 Expansion of a sensor chamber from a small inlet. Whatlimits how fast the eel’s olfaction chamber in Fig. 26 expandarea downstream of the tiny anterior naris? Lighthillf7g first con-sidered this issue in terms of the branching of the arteriesbronchial tubes. Briefly, the area increase slows the flowraises the static pressure, so except at low Reynolds numberis a danger of flow separation. Lighthill’s rule-of-thumb to avthis allows no more than a 20% increase in flow cross-sectarea per channel bifurcation. At least in the arteries, nature sto obey this rule. Cross-sectional area data are not availabthe eel’s olfactory chamber, but are available for the compacase of a beagle’s nosef64g. These data show that the maxillotbinate region expands about linearly in cross section to stimes its initial area over a downstream distance of only 35According to Lighthill’s rule, a bifurcation should occur initiaevery mm or so to prevent separation. This is hard to checkthe available data, but it is clear that the maxilloturbinates burwith branching complexity in this region. Thus separation pretion goes hand-in-hand with the need for large heat-exchasurface area in the beagle’s maxilloturbinates.

5.3.4 Heat and mass transfer analogy. Osborne Reynolds1842–1912d recognized that mass, heat, and momentumtransferred by similar physical mechanismssmolecular motion inlaminar flow, eddy motion in turbulent flowd. The analogy be

Fig. 26 The path of water through the olfaction chamber of aneel, top view, redrawn from illustrations in †40,253,325‡

tween these fluxes—Reynolds analogy—is useful in sensor cha

206 / Vol. 127, MARCH 2005

heo-rs

e,ar-

n

menal,a-stle,

eor-

r-oftu-s-o

n

nddere

almsforle

en.

thn-er

re

ber design, where a high mass transfer rate of odorant moleto olfactory surfaces is desired. Concomitant momentum traby friction is required by the analogy, and heat transfer mayoccur. Thus odorant mass transfer is related to the fluid prequired to overcome sensor-chamber friction losses. The emcal relations describing Reynolds analogy are well-knownreadily available, e.g.,f91,327,328g, and do not bear repeatihere.

The utility of Reynolds analogy in sniffing for chemical trawas first recognized by Fraim et al.f311g, who designed an eplosives preconcentrator based on heat exchanger technolomodern microelectronics, strong thermal generation occurssmall confined space and must be removed by a fluid flow.analogy with an olfactory sensor chamber is obvious fromabove, so we can learn design principles from the volumeshave been written on electronics cooling, e.g.,f21,22,327,329g.

Finned heat exchangers expose large surface areas to thfor efficient heat transfer, a principle already in use in natuthe nasal turbinates, as well as in the air-conditioning induf91g. One must make sure the flow is along the fin channelsnot across them, or the process becomes stalled. For flow inshaped ducts the Reynolds number is based on the hydrauameterDh=4A/P, whereA is the cross-sectional area andP is theduct perimeter. Loss coefficients are known for many situate.g., discharging the flow abruptly from a small tube into a lchamberswhich dissipates one velocity head, Sec. 2.3d. Loss for-mulas are likewise known for screens, grids, protuberancesrough channel walls.

In particular, the stacked eel lamellae of Fig. 26 are strikisimilar to the finned heat sinks used to cool electronicsf21g. Heat-sink design principles encourage low flow speeds to minimizpressure loss for a given performance level. Straight finsshown to outperform other fin shapesse.g., cylindersd f327g. Finsand lamellae should also be aerodynamic, with rounded leedges and tapered trailing edges.

6 Homeland Security ApplicationsDown through the ages, civilization only flourished in encla

sufficiently well-defended to prevent the ravages of barbarThe terms “homeland security” and “terrorism” were not popthen, but the results of a security failure were nonetheless dtating. Fluid mechanics has been important in homeland seall along, providing standoff barriers against invasion, weatechnologies, mobility, etc.f330g.

Many believe that the modern solution to the asymmetry oterrorist threat lies in technology. Here we discuss a piece otechnology that concerns gaining information by detecting chcal traces in the air or water. Like the prey animals discuearlier, we need constant environmental awareness but,them, we cannot sleep standing up.

Homeland security thus requires continuous environmmonitoring. The unusual, however minor, can be a crucial warof an impending attack. For example Carranza et al.f331g, whilemonitoring the atmosphere for heavy metal traces, noticed anesium spike on the July 4th holiday—not terrorism that timeit could have been.

6.1 Canine Detection.So much has been written about traing dogs for chemical trace detection that it serves presenposes merely to cite some key references and give acommentary.

Much of the scientific basis of canine detection was learecently at the Auburn University College of Veterinary MedicAuburn, Alabamaf66,332,214–217g. There, operant conditioninexperiments yielded sensitivity curves like those of Fig. 27,quantify the threshold levels of specific trace chemicals thatcan detect. These data also set the sensitivity standard for arnoses, a standard that has already been exceeded in at le
m-casef292g.

thres

ive

nr ey tashe

ithingfing

itseeanhelog

tim

deo

nit wanex

uffean

liti-rd

nifiAir,poxp, aan

ingefft atinctitra

hichuctionh anrthe

d ande alsoay ofo in-

sam-sea

rticlenxplo-

ote

of airted toughignifi-lated

racescanis-

nal,

byion.ector.is thet,races

r way.tandflows

allof then of

founde andaitoes.iffersts arenualn in

pleen fur-to theim-ceed inmentsheat

a-

ctingm

st takee hu-

Extraneous odors present while sniffing can interfere widog’s ability to detect a scent and can therefore raise the tholds shown in Fig. 27, but to do so requires a comparatenormous extraneous odorant levelf217g.

Dogs respond to the most-volatile compounds present iexplosive, not necessarily to the explosive species itself. Foample in detecting C-4 plastic explosive, the dog is not likelrespond directly to the explosive component RDX, which hvery low vapor pressure, but rather to compounds like cycloanone, a solvent used in RDX productionf215,216,333g.

Gazit and Terkel reportf334g that heavy exercise interferes wa dog’s olfactory ability, since the dog cannot sniff while pantThey also use a microphone to pick up the sound of snifwhich provides useful feedback to the dog handlerf335g.

While handlers have commented on the most desirable tratheir dogsf336g, the desirable traits of handlers appear to nmore workf332,337g. Being the brains of the team, the dog hdler has a responsibility to be aware of the issues discussedsuch as human skin flakes, chemical plumes, micrometeoroand the aerodynamics of air and ground sniffing. Syrotuckf125gshould be required reading. Thermal plumes can also somebe visible to the unaided eye of the handlerf46,48,338g. Thesefluid-dynamic issues might well be illustrated in a training vifor dog handlers.

Despite the issues of expense and limited duty cycle, catrace detection is a trusted mainstay of homeland security thanot soon be replaced by artificial nose technology. FurtonMyers f333g found that detector dogs still represent the bestplosive detection means available, since artificial sniffers sfrom poor sampling systems, interference from masking odorand limited mobility.

6.2 Cargo Screening.Cargo screening has become pocized because of the high cargo volume crossing national boand the extreme measures that are needed to screen a sigfraction of it for explosive, chemical, or biological weapons.sea, rail, and truck cargo are all at risk. Methods that rely uradiation to probe cargo containers are cumbersome and esive, detection dogs do not have access to cargo interiorsmanual unloading for inspection requires far too much timelabor to be practical.

Still, sniffing the air inside cargo containers—without openthem—for trace contraband is an approach that can be bothtive and affordable, and there is a precedent for it. Fine ef339g, for example, patented a method to do this by injeccompressed air into a cargo container, applying suction, collethe sampled air in a hood, and interrogating its contents for

Fig. 27 Canine olfaction sensitivity curves measured at Au-burn University, redrawn from http://www.vetmed.auburn.edu/ibds/. Except for methyl benzoate, a cocaine derivative, alltrace chemicals shown are explosive-related

chemicals.


ah-

ly

anx-oax-

.,

ofd-re,y,

es

neilld-r

ts,

erscant

nen-ndd

ec-l.

gngce

Another example is the MDS Sciex CONDOR system, wsearched cargo containers for explosive traces using a sprobe fitted with a brush that fed an adsorption cartridge witairflow flow rate of 25 liters/sf340g. The interior of containers othe exterior of cargo pallets was sampled by contact withbrush for about 30 s, whereupon the cartridge was removedesorbed into a mass spectrometer. Cargo containers wersampled through their ventilator ports or through a 10 cm arrdrilled holes that ended up not being acceptable to the cargdustryf302g. This substantial system employed a 9 m heatedpling line and applied mechanical agitation even to full-sizedcontainers. Inside the container they found a gradient of padispersion leading up to the sourcef302g. Trace contaminatiowas even detected in some containers not holding any bulk esives: Once upon a time Kilroy had been there.

Still another approach to cargo sampling is the British RemAir Sampling for Canine OlfactionsRASCOd systemf341,342g. Aprobe inserted into a cargo container pumps out a volumethrough a polymer mesh sample canister that is later presena detection dog for analysis. Sixty liters/min are drawn throthe canister for 2 min. Success appears to depend upon a scant explosive vapor headspace inside the container. A regeneral discussion of collecting and transporting chemical tin evacuated flasks, canisters, sampling bags, and sorbentters is given inf306g. If you cannot get the proboscis to the sigtake the signal to the proboscis.

Jenkins f343g patented a method to interrogate baggageplacing it in a compartment provided with vacuum and vibratAirborne chemical traces thus liberated are drawn into a detA recent adaptation of this approach to cargo trace detectionRay Detection Discovery CERT systemf344g, which uses heapressurization, vibration, and gas jets to dislodge explosive tfrom palletized air cargo in an airtight enclosure.

Finally, research on cargo trace detection is currently undeat Penn State, University Park, PAf345g, funded by the U.STransportation Security Administration. The goal is to understhe internal airflows of cargo containers and to sample thesefor trace detection without opening the containers.

6.3 People Sampling.People sampling is a unique topic inof chemical trace detection. Aircraft passengers are awaredanger of terrorism, but are also very sensitive to the invasiotheir personal space. A NRC study of passenger screeningthat there is a strong relationship between public acceptancthe perception of riskf346g. It nonetheless irks passengers to win long lines, parcel out their belongings, and take off their shBeing sniffed by a dog is particularly offensive. People-sndevelopers need to be aware of these issues, just as terrorisurely aware of them. But despite all this, the total U.S. anpeople-screening outlay is expected to grow from $590 millio2001 to $9.9 billion in 2010f347g.

Hallowell f348g reviewed the available technology for peoscreening in 2001. Since then several approaches have sether development. One study recommended that research in“vapor space” surrounding a suicide bomber might lead toproved means of detectionf48g. Misconceptions about that spaand its true nature—the human thermal plume—were describSec. 3.2.1. Here we discuss recent people-sampling developin light of what is now known about the aerodynamics andtransfer around people.

6.3.1 Human olfactory signature. The human olfactory signture was pioneered as a topic of study by Dravnieksf314,349g.Human scent arises from bioeffluents and from bacteria aupon the skin and its secretionsf125g. People also carry with thean olfactory image of their recent environmentf222,314g. Anysuccessful scheme to detect trace chemicals on people muaccount of this olfactory signature and its conveyance by thman thermal plume and wake.

Dravnieks developed a “body tube” sampler to acquire chemi-

MARCH 2005, Vol. 127 / 207

te-on

n in-log

loythic

plod

ctohin

g

cluairctlowrstra

n2.5pu

s ii-deedl.

anewthT

s ttiof thororth

flo

suaym

ducoistennairlea

ritthelumd

thee u

uph

Statea-

pling

for

jecte.teloth-%.l, but

bothnity

portmal

ven-e of

ortal,sent

heound-Thisd the-d

isnarys

ssion

n thetion

peed

cal trace signals from peoplef349g. Bethune et al.f350g and Mayand Pomeroyf126g used similar “dispersal chambers” for bacriological studies of human subjects. Such enclosures arepractical for people-sampling in the lab, but they have beestrumental in contamination studiesf351,352g and in aviation security research leading up to the present portal technof123,124,353,354g.

There is anecdotal evidence that artificial noses were depto detect troops in the Vietnam jungle, and more recently incaves of Afghanistan. Building upon genetic research with mthe current DARPA Unique Signatures program seeks an exable chemosignal corresponding to an individual’s geneticallytermined odor type. If found, it will be used to develop a detefor the presence of “high-level-of-interest individuals witgroups of enemy combatants.”

6.3.2 Portals for screening people. Portals for screeninpeople have generated many patents, e.g.,f355–358g, but only afew practical devices. Some of the approaches that failed in“wind-tunnel” portals that moved massive amounts of“saloon-door” portals that required physical contact with subjeand “telephone-booth” portals that were too confining or too sDuring the development of these devices a goal of 6 s/pesampling time was set by the U.S. Federal Aviation Administion sFAAd, but was difficult to achieve.

General reviews of people-screening portals are givef348,310g. The role of air jets was described earlier in Sec.and related material on explosive trace standards and comtions of the human thermal plume is found inf130,359–363g.

The first published investigation of portal aerodynamicprobably Schellenbaumf364g, who examined various airflow drections in tube-type and booth-type portals. He usedepartment-store mannequinssee Sec. 3.2.2d, so his results takno account of the human thermal plume. He recommenddownward airflow over the subject with collection at foot leve

Hobbes and Condef365g studied the aerodynamics of“open-clamshell”-type portal by CFD simulation. This portal dra large flux of air—350 liters/s for 10 s—horizontally acrosshuman subject and into inlets for subsequent interrogation.induced airspeed was up to 2.4 m/s, more than ten timecommonly accepted maximum for comfortable room ventilaf366g. Vortical recirculation was generated on the lee side osubject. Air-jet puffer effects and the release of explosive vapa passive contaminant were also simulated. Although this pdesign lacked sophistication, it was the first study to applytools of fluid dynamics to people sampling.

Sandia National Laboratories developed a vertical-downportal in the 1990sf356,368,367,358g, a version of which is nowcommercialized as the Smiths Detection Sentinel II. A humanject enters the portal and turns 90 deg, whereupon an arrair-jet puffers and two ceiling slot-jets sweep the body for checal traces. The cross-sectional area of the open portal is renear the floor, where a portion of the induced downflow islected for analysis by an IMS detector. This is purely jet-asssniffing, since thermal convection from the human subject caplay a significant role. It bears a similarity to air-shower ordouche devices used for particle removal from clothing in croom practicesf17g Chap. 14, andf369gd.

Quite a different approach is taken in the GE SecuEntryScan3, Fig. 28, which uses the natural upward motion ofhuman thermal plume to collect and preconcentrate the poverhead for IMS interrogationf357g. This is convection-assistesniffing, but air-jet puffers also play a role in both rufflingsubject’s clothing to release particles and in encouraging thward airflow. An aerodynamic contractionssee, e.g.,f370gd is usedoverhead to capture the human thermal plume at a flow rate50 liters/s and pass it through a 10310 cm metal-mes
preconcentrator.
208 / Vol. 127, MARCH 2005

ly-

y

edee,it-e-r

de,s,.

on-

in,ta-

s

a

a

eheheneastale

w

b-of

i-cedl-dot-n

y

e

p-

to

Several years of research on this portal at Pennf83,124,353,354,371g have yielded the following results, summrized briefly:

1. The human thermal plume is a natural whole-body samsystem, from the toes to the top of the head.

2. As shown in Fig. 4,brief puffer-jet impingementse.g.,50 msd is the principal means of liberating trace particlesdetection.

3. Most of the readily-available signal from a human subcan be had in a few seconds of overhead sampling tim

4. The detectability of 10mm-range explosive particulatraces originating beneath the clothing is affected by cing porosity only when the porosity is less than about 4

5. Layered clothing reduces the chemical trace signal levethe signal remains measurable.

6. Trace explosive sources beneath the clothing transfer tothe clothing and the skin, but remain highest in the viciof the source.

7. Human skin flakes do not figure prominently in the transof dry crystalline explosive traces in the human therplume.

8. Lateral air currents in the range of normally acceptabletilation drafts do not substantially affect the performancan open portal such as the one tested here.

9. When pure explosive vapor was released inside the pup to 25% of it was captured and detected by the premetal-mesh pre-concentrator.

6.3.3 Human aerodynamic wake. For a lateral airspeedUsor walking speed in still aird of more than about 0.2 m/s thuman thermal plume ceases to exist. Instead, the thermal bary layer of a person is swept downstream to form a wake.can be seen by comparing the Reynolds number Re anGrashof number Gr=gbDTL3/n2, whereb is the volumetric thermal expansion coefficient,DT is the temperature difference, anLis the characteristic width of the human bodyf372g. Without lat-eral airspeedU, Re→0 and Gr/Re2@1, meaning that the flowdominated by free convection. On the other hand, for ordiwalking speeds in theU=1 m/s range, Gr/Re2!1 and the flow idominated by forced rather than free convection. SinceDT nolonger matters in the latter flow regime, we use the expre“human aerodynamic wake.”

Several investigators have studied lateral airflow effects oflow about stationary humans, mostly motivated by air pollucontrol and the reduction of worker exposuref135,373–377g. In adifferent reference frame, however, a person walking with s

Fig. 28 „a… The author standing in a prototype version of theGE Infrastructure EntryScan 3 explosive detection portal „PennState photo by Greg Grieco … and „b… schlieren image of portalwith subject „L.J. Dodson … and with air-jet puffers firing

U in still air produces an aerodynamic wake in which the wake


, foth

ntrso

empldescta, e

ingde

r-vor

uausthcuwaheshhow

quodak

r coy dix

eare th

age-muchvere

se-weaktime

an

evenog’sosespedfar.

in

plo-l levelineen-

te toturer the2,4-thena-sive

d soil

flow follows the person. The human chemical trace signaturemerly in the human boundary layer, becomes distributed inwake. It is common experience that a walking person’s scestill air is noticeable to a fixed observer only after that pewalks by.

This has been exploited as a means of “sniffing” human chcal traces by sampling the wakes of walking peof372,378,379g. The advantage over the portal technologyscribed above is that people are already walking in securitynarios such as airports and, by not making them stop and sthe sampling time can be reduced substantially. For exampleperiments f378g have shown that the wakes of two walkpeople, one following the other, are aerodynamically indepenfor separation times greater than about 2 s.

The Penn State workf372,378,379g, still in progress, has chaacterized the human aerodynamic wake as the unsteadyshedding from an irregular 3D cylindrical bodysFig. 29d. Theirregularity mostly concerns the legs, which act as individshedding cylinders with through-flow between them. This cathe wake to dissipate more rapidly downstream of the legsdownstream of the torso, where a large recirculation zone ocThere is also a downwash component that causes the lowerto interact with the ground and to spread laterally. Despite tseveral complicating aspects, the human wake neverthelessunmistakable Kármán vortex shedding in the dorsal plane sin Figs. 29sbd and 30sbd.

A CFD solution of the Reynolds-averaged Navier-Stokes etionsf372g, despite a simplified representation of the human bis able to portray all the significant aspects of the human wflow that are seen in the experiments. For example, a scalataminant representing a chemical trace from the human bodcays exponentially with distance downstream as the wake mout with the surrounding air. Thus one should sample thewake just downstream of a walking person in order to acquirmaximum trace signal strength.

Fig. 29 Diagrams of motion in the human aerodynamic wakefrom smoke flow visualization experiments †372,378‡, „a… me-dian plane and „b… dorsal plane

The design of a portal to sample the human aerodynamic wa


r-isinn

i-e-e-nd,x-

nt

tex

lesanrs.ke

seows

n

a-y,en-e-eslye

was first suggested inf357g. It appears possible to take advantof the wake’s momentum in this regardf378,379g. However, compared to the case of the human thermal plume, sampling alarger volume of air over a shorter time interval will be a setest of pre-concentrator technology.

6.4 Landmine Detection.Landmine detection is anothervere test for sniffers, as described in Sec. 3.3.2, because oftrace signals outdoors in the weather and terrain. At the samethe global landmine crisisf380–383g extracts a shocking humand economic price and demands solutions.

Dogs are the principal detectors of buried landmines,though it puts them and their handlers at risk. The DARPA DNose program in the late 1990s sought to produce artificial nof similar sensitivity, and several of such were actually develof287,288,292,384g, though none has supplanted the dog soYinon f310g and Pamulaf384g have both surveyed this topicmore detail than present space allows.

Briefly, as part of the mentioned DARPA program an Exsives Fate and Transport Team characterized the trace signathat one can expect to find above a buried landmf162,385–387g. Trace explosives and related mine compoundster the surrounding soil, collect below the mine, and percolathe surface in a “halo” pattern about the mine location. Moisand daily variations in soil temperature aid this process. Foubiquitous 2,4,6-TNT mine, the prevalent trace signal is fromDNT, which is more volatile than TNT and is present inmilitary-grade explosive as a contaminant. It is this “odor sigture” that a dog actually detects, rather than the bulk explof215g.

The mass of explosive-related compounds in contaminate

Fig. 30 Corresponding to Fig. 29, computed instantaneousmotion of a scalar contaminant in the aerodynamic wake of asimulated human †372‡, „a… median plane and „b… dorsal plane

keabove a buried mine was found to be on the order of tens of

MARCH 2005, Vol. 127 / 209

th, builis a

iveen5.rryal.emd

pt tac

-ert t

-pliowesdrthosor-

iteinaev

ouse-fillen.se

n tnvfi

pleacctiesm

u-iseristr

ninrance oha

tiot a”

r

uredi-

talsand

ch-trace

l ol-am-fec-

tract

isease

anyi, andn is

real-ficultsamence isthanrrentplace-urrent

es,Theyines,d pipe-mustmen-erous.ctionation, andpro-ce,

n are

hicleFind-and

micrograms/kilogram of soil. 2,4-DNT vapor is also present inboundary layer directly above a typical anti-personnel mineonly at a concentration of perhaps 100 picograms/ liter at eqrium under ideal atmospheric conditions, and such conditionnot usualf387g.

Overall, then, the chances are bleak of sniffing explosrelated vapor in the air above a buried mine. Anecdotal evidfrom dog handlers that their dogs air-scented landmines atmust describe extraordinary rather than usual circumstances

Soil particles, though, are another matter, since they cathousand-fold higher concentration of explosive-related signdog’s ability to stir up, inspire, and probably desorb trace chcals from such particles was discussed in Sec. 4.3.1 anf103,166g. Some developers of landmine sniffers also attemtake advantage of soil particles by thermally desorbing trfrom the soil using laser radiation before sniffingf388g, by dis-turbing loose particles using air jetsf310,384g or ultrasonicsf389g, or by collecting particles electrostaticallyf390g.

Finally, Mineseeker.comf391g floats a mini blimp over a minefield and uses radar to locate landmines. One could lowelephantlike sampling trunk with assisting air jets almosground level and sniff for them as well.

6.5 Biohazard Detection.The topic of sniffing for bioaerosols was introduced in Sec. 3.1, including some unique samand detection problems and the issue of distinguishing biagents from the natural airborne background. Chemical plumone way of dispersing biowar agents—were likewise covereSec. 3.3.3 and aerobiology was introduced in Sec. 1.1. Fureferences are also available on sampling biological aerf106,108g, biosensorsf392g, and chemical and biological terroism f12g.

Consider the anthrax letters that were mailed in the UnStates in late 2001. Finely powdered anthrax spores in ordpaper envelopes became airborne due to handling, costing slives and contaminating two large mail-handling facilitiesf393g.Postal delivery-bar-code sorters squeezed the air and sporesthe envelopes by belts and rollers. 1mm–3 mm anthrax sporecan pass easily through the 20mm pores of a typical paper envlope even if it remains unopened. Tearing open an anthrax-envelope was found to cause massive airborne contaminatio

The Northrop-Grumman Corporation, Baltimore, MD, subquently developed a biohazard detection systemsBDSd for theU.S. Postal Service to detect any such future attacks upopostal system. BDS compresses mail to force the air out of elopes and captures it in a sampling hood positioned over thepinch point in the mail processing operation. From the samairflow a Spincon high-volume liquid-based cyclone collectorquires biological particles and uses the polymerase chain reaand DNA matching to detect the trace presence of such specBacillus anthracisin the mail. BDS is a fully automated systethat sounds an alert when a biohazard is detected.

7 Other Applications

7.1 Medical Diagnosis.Since the beginning of medicine, hman body and breath odors have been used to diagnose df394,395g. Dozens of diseases are known by their characteodors, such as acetone breath from diabetesf396g, the putrid odoof scurvy, and the sour odor that precedes mononucleosisf397g.Also included are characteristic odors due to chemical poisoand substance abuse, and even schizophreniaf398g. The populapress has made much of pet dogs purportedly sniffing out cin their owners. However, lacking quantification and in the facmodern laboratory analysis, olfaction in medical diagnosislanguished.

One reason for this is that most methods of artificial olfacfor medical diagnosis are intrusive and slow. Dravnieks ef314,349g and Distelheim and Dravnieksf399g used a “body tube
sampler sSec. 6.3.1d and gas chromatographsGCd-mass-
210 / Vol. 127, MARCH 2005

eutb-re

-cem

aAi-inoes

ano

ngar—inerls

dryeral

t of

d

-

hee-

rstd-ion

as

aseic

g

erfs

nl.

spectrometric analysis. Cotton swabs, breath bagsf400,401g, andskin headspace collectorsf402,403g all work without the need foactual sniffing.

Gardner and Bartlettf9g speculated that “Perhaps in the futthe electronic nose will be able to sniff the human bodyrectly…” That future has already arrived in the form of porthat sample the human thermal plume, Sec. 6.3f83,120,124,357,371g. However, modern nonintrusive portal tenology has yet to be applied outside the realm of explosivedetection for security screening.

Nonetheless there is considerable recent activity in artificiafaction for medical diagnosis, mostly using only intrusive spling methods. Applications include staph, strep, and e-coli intions, breath signals for uremia and liver cirrhosis, urinaryinfections, tuberculosis, and airway inflammationf9,404,405g.Continuous e-nose monitoring of disease stages and global dsurveillance have been suggestedf406g but not yet realized.

Clearly, modern sniffing for medical diagnosis still has mhurdles to overcome, as described by Persaud, PisanellEvans f407g. Despite its enormous potential, this applicatiopresently more driven by sensor research and developmentf392gthan by sampling issues, and there is not yet a commercialization even of a breath analyzer. The medical arena is a difone; human variations can produce different odors from thedisease, misdiagnoses are bad news, and chemical interferecommon. Also the funding tends to favor therapeutic ratherdiagnostic instrument development. Finally, even though culab analyses of blood samples and cultures take days, any rement technology must at least achieve the accuracy of the cmethods.

7.2 Leak Detection.Wand-type leak-detector sniffer probFig. 31 f408g, were discussed in Sec. 3.2.4 and elsewhere.have very broad industrial applications in leak-testing turbcondensers, heat exchangers, tanks, refrigerant systems, anlines. Explosive rocket propellant leaks must be found, asleaking solvents or fuels that pose fire, explosion, or environtal hazards. Natural gas leaks are expensive as well as dang

Usually a hand-held sniffer probe acquires a sample by suand conveys it to a sensor such as a helium or flame-ionizdetector. The sampling time rises rapidly with the hose lengthfor long hoses of typical 1-cm-or-less diameter it becomeshibitive. Example calculations for sensitivity, standoff distanlinear speed of probe motion, response time, and calibratiogiven in f139g.

Some other sniffer types are not hand held, but rather vemounted. In one case, Boreal Laser’s aircraft-mounted GaserAB system is flown over natural gas pipelines, sampling

Fig. 31 „a… Schlieren image of an acetone vapor leak cascad-ing downward and being inhaled „arrow … by the snout of theCogniscent artificial nose „b… a hand-held, wand-type snifferprobe is used to search for a natural gas leak, visualized bylarge-field schlieren optics †46,408,411‡

analyzing the air for leaks. Dogs are also trained to sniff for gas


arroan

leaykinignllth

r-rso

ur

amc-

let

ting

stilt th

nd

ceosrti

rnxancal

atesl

foont a

or

noroair

lateceth

recon

ain

pi

is tuseogatin

sen

little

noldsnae

t all,Thed to

adehisti-im-

re as

turalad.

ed, at

pleelse

urity

s as-ew”

n-n the

pos-tural

micary

n.par-u-

tion

nts,hu-eeze.f the

, theten-tion.and

cal-nd

ction.tion

sorb-, as

fortorycal

y to

centg for

leaks.Hand-held sniffer probes are accurate and sensitive, but se

ing is slow and coverage of large areas is labor intensive. Bfield standoff optical methods can supplement probe detectiocovering large areas quickly at reduced sensitivity levels. Atthree such approaches have been suggested:s1d IR thermographf409g, which depends upon a thermal difference between leagas and the ambient air and usually only examines ground stures;s2d Schlieren imagingf46,410,411g, which requires a smarefractive-index difference but images the air itself rather thanground; ands3d laser absorptionf412g, which can detect isothemal airborne leak plumes but requires expensive and cumbeequipment.

8 Summary and Future Directions

8.1 Summary. For convenience, the high points of this toof the world of sniffers are repeated in list form:

1. Little previous effort has been spent on the biofluid dynics of sniffing or on the design of sniffers for artificial olfation.

2. Nature favors energy-efficient faired bellmouth-type inrather than sharp-edged holes.

3. Brief air-jet impingement is the principal means of liberatrace particles from surfaces for detection by sniffing.

4. The proper model for airflow about the human body inair is the human thermal boundary layer and plume, no“personal activity cloud.”

5. Plumes bearing chemical traces can be classified and ustood based on their buoyancy and momentum.

6. Nature teaches proximity sniffing, but the most advannatural nostrils, such as the variable-geometry dog’s nalso use exhaled air jets to enhance the reach and pauptake of the process.

7. A vertebrate does not require variable-geometry extenares in order to be macrosmatic, but several of those eined here do have them. We know something of their fution in dogs, but almost nothing in the case of other anim

8. The natural narial sampling orifice of terrestrial vertebris typically either round, inverse-comma shaped, orshaped.

9. A proboscis, nostril, and mobile platform are requiredvertebrate sniffing, but proboscis length is only weakly cstrained by nature, as demonstrated by the elephankiwi.

10. Small turbomachines can move fluids for olfaction meffectively than nature’s usual bellows action.

11. Macrosmatic terrestrial vertebrates usually have largetrils for high airflow rates and long straight snouts to pvide enough space for both the olfactory and theconditioning apparatus.

12. None of the animals can afford the time to collect anddesorb odorants. They acquire the odorant by sensorgenerating neuronal signals that are sent directly tobrain in real time.

13. The peak mammalian sniff rate of a few hertz ignohigher-frequency information, such as some of thattained in chemical plumes. This suggests that a man-mcontinuously-inhaling sniffer could gain an advantagefrequency response by not having to accommodate restion.

14. For olfactory sensor chamber design the best modelfish: simple, elegant, and not complicated by dualMounting the sensor chamber off to the side, as in the dnose, appears just to be nature’s way of accommodolfaction with respiration.

15. Reynolds analogy plays an important role in olfactory
sor chamber design.

ch-d-byst

ga-

e

me

-

s

le

er-

de,cle

alm--

s.sit

r-nd

e

s---

rllse

s-

de

ra-

he.’sg

-

16. Mankind’s worst enemy, the mosquito, has receivedattention from fluid dynamicists.

17. Invertebrates who track chemical plumes use the Reynumber to control the effective porosity of their antensensor hairs during sniffing.

18. We can understand a lot about natural sniffers but nobecause biomimicry is a sort of reverse engineering:final product is available, but no logical path is revealeget there.

19. Compared to nature, the sniffers now used in man-mdetectors and artificial noses are almost devoid of sopcation. Several possibilities are suggested for theirprovement.

20. Vapor and particle sampling are both important in natuwell as in artificial olfaction.

21. In people sniffing, the human thermal plume is a nawhole-body sampler, from the toes to the top of the he

22. The human aerodynamic wake has been characterizleast initially, by both experiments and computations.

23. Modern nonintrusive portal technology for sniffing peohas yet to be applied to medical diagnosis or anythingoutside the realm of explosive trace detection for secscreening.

24. This paper has collected scattered information and hasociated diverse fields in the hope of establishing a “ntopic in biofluid dynamics.

8.2 Future Directions. Future directions for the further uderstanding and development of sniffers are summarized ifollowing list:

1. Study the nostril aerodynamics of the rabbit, rat, and osum in order to broaden the present knowledge of nasniffing systems.

2. Carry out an experimental and computational fluid-dynastudy of an eel’s or shark’s olfactory capsule as a primexample for sensor chamber design in artificial olfactio

3. Study the internal aerodynamics of the dog’s nose byticle image velocimetrysPIVd measurements and CFD simlation as another important example for artificial olfacsensor chamber design.

4. Carry out flow visualization, PIV velocity measuremeand a CFD simulation of the airflow around the upperman body, breathing zone, exhaled jets, cough, and sn

5. Perform turbulent and unsteady-flow CFD simulations oflow inside the human and dog’s nose.

6. Gain a better understanding of mosquito aerodynamicseffects of ambient wind, mosquito plume tracing, and potial aerodynamic strategies to defeat mosquito host loca

7. Beginning at the fundamental level, examine currentprospective new artificial sniffer types in terms of practiity, efficiency, proper use of fluid-dynamic principles, athe performance criteria stated in Sec. 5.1.3.

8. Conduct research to better understand jet-assisted olfa9. Study more effective means of aerodynamic sniffer isola

from the effects of an ambient breeze.10. Develop ways to replace the present adsorb-de

analyze-detect cycle with more direct olfactory sensingin nature.

11. Develop flexible, controllable “elephant trunk” snoutsland-based robot sniffers in order to improve their olfacmobility and performance by aiming their trunks at loregions of interest.

12. Apply modern nonintrusive portal sampling technologhuman medical diagnosis.

13. Produce an educational film on the fluid mechanics of splumes, skin flakes, and the aerodynamics of scentin
the dog handler and the artificial nose user.
MARCH 2005, Vol. 127 / 211

ilam

f

,wiitio

ewso

veriffeineic dheeathis

e fperar. S, Nr. ff iuthllerewistrPAt t

n in

,

,

ctor

y-Me

d

20

ews

e-

n

OH

Pa

, anIn-

col.

w

n

ow

thepts

13–

ed.,

.n

94.

n Der

yclic

e, G.an

ured,” J.

tion

tromeshe

ngol.

000,nsor

is ofcentra-

entila-

Fische.

CrabsCurr.

on of

ues in

rsity

tionse,” J.

Phe-

ternaleer-

tentialuncil

-

dt, F.on

. T.,etryicol.

Finally, various studies and working groups have made simsuggestions for future research, including more study of mamlian olfaction, sampling, and preconcentrationf205g, biomimeticsensing focused on distributed low-cost sensorsf48g, factories othe future filled with cheap tiny olfactory and other sensorsf413g,rapid wristwatch-sized biodetectorsf414g, low-power, distributedunattended, miniaturized remote sensors, robotic “insects”onboard sampling and sensors, and rat packs with global posing system transponders trained to detect explosives in city sf48g. Some of this speculation is far fetched, but we can reaably expect that sniffers will eventually no longer bead hocaf-terthoughts in artificial olfaction, as they are now in almost ecase. Small, portable, inexpensive e-noses with integral snwill eventually become common useful tools, although tracanine detectors will not soon become obsolete. The genettectives, e.g.,f237,415g, will also have more to say about torigin and function of the various noses reviewed here. Mwhile fluid dynamicists have a lot of interesting work to do ontopic.

AcknowledgmentsThe author thanks the ASME Freeman Scholar Committe

providing him the opportunity to write this review. The pabenefited especially from critiques by S. Vogel, D. M. Stoddand L. R. Witmer. The author also thanks B. A. Craven, JKauer, A. Kepecs, M. J. Kennett, P. Mielle, E. G. PatersonSobel, and J. E. White for helpful discussions, T. S. Denny, Jthe dog’s nose cast, and the several gracious contributors oages who are acknowledged in the figure descriptions. The afinally thanks his laboratory staff, L. J. Dodson and J. D. Miand his students for their support. Some elements of the reviresearch were supported by the US Federal Aviation Admintion, the Transportation Security Administration, and DAROpinions expressed in this paper do not necessarily reflecpolicies of these sponsoring agencies.

Referencesf1g Eisenberg, J. F., and Kleiman, D. G., 1972, “Olfactory Communicatio

Mammals,” Annu. Rev. Ecol. Syst.,3, pp. 1–32.f2g Zwaardemaker, H., 1895,Die Physiologie des Geruchs, W. Engelmann

Leipzig, Germany.f3g Stoddart, D. M., 1980,The Ecology of Vertebrate Olfaction, Chapman & Hall

New York, pp. 1–33.f4g Axel, R., 1995, “The Molecular Logic of Smell,” Sci. Am.,273s4d, pp. 154–

159.f5g Doty, R. L., 2001, “Olfaction,” Annu. Rev. Psychol.,52, pp. 423–452.f6g Kauer, J. S., and White, J. E., 2001, “Imaging and Coding in the Olfa

System,” Annu. Rev. Neurosci.,24, pp. 963–979.f7g Lighthill, M. J., 1975,Mathematical Biofluiddynamics, SIAM, Philadelphia.f8g Weinbaum, S., 1996, “Biofluid Mechanics,” inResearch Trends in Fluid D

namics: Report from US National Commission on Theoretical & Appliedchanics, J. L. Lumley, ed., Springer, New York, pp. 296–309.

f9g Gardner, J. W., and Bartlett, P. N., 1999,Electronic Noses—Principles anApplications, Oxford University Press, Oxford.

f10g Pearce, T. C., Schiffman, S. S., Nagle, H. T., and Gardner, J. W., eds.,Handbook of Machine Olfaction, Wiley, New York.

f11g Rouhi, A. M., 1997, “Detecting Illegal Substances,” Chem. Eng. N75s39d, pp. 24–29.

f12g National Research Council, 1999,Chemical and Biological Terrorism: Rsearch and Development to Improve Civilian Medical Response, NationalAcademy Press, Washington, DC.

f13g Watson, J. G., and Chow, J. C., 2001, “Ambient Air Sampling,” inAerosolMeasurement: Principles, Techniques, and Applications, 2nd Ed., P. A. Baroand K. Willeke, eds., Wiley, New York, pp. 821–844.

f14g Monteith, L. E., and Rubow, K. L., 2001, “Air Movers and Samplers,” inAirSampling Instruments for Evaluation of Atmospheric Contaminants, 9th Ed.American Conference of Governmental Industrial Hygienists, Cincinnati,pp. 233–280.

f15g Singh, M., and Sioutas, C., 2003, “Assessment of Exposure to Airborneticles,” in Indoor Environment: Airborne Particles and Settled Dust, L.Morowska and T. Salthammer, eds., Wiley, New York, pp. 374–386.

f16g Subramaniam, R. P., Richardson, R. B., Morgan, K. T., Kimbell, J. S.Guilmette, R. A., 1998, “Computational Fluid Dynamics Simulations ofspiratory Airflow in the Human Nose and Nasopharynx,” Inhalation Toxi
10s5d, pp. 473–502.
212 / Vol. 127, MARCH 2005

ra-

thn-

ersn-

yrsde-

n-

or

t,..

orm-or

,eda-.he

y

-

02,

,

,

r-

d

,

f17g Baturin, V. V., 1972,Fundamentals of Industrial Ventilation, Pergamon, NeYork, pp. 12–24.

f18g Heinsohn, R. J., 1991,Industrial Ventilation, Wiley, New York.f19g Heinsohn, R. J., and Cimbala, J. M., 2002, “Ideal Flow,” inIndoor Air Quality

Engineering, Marcel Dekker, New York, pp. 517–580.f20g Goodfellow, H. D., and Tähti, E., eds., 2001,Industrial Ventilation Desig

Guidebook, Elsevier, New York.f21g Kordyban, T., 1998,Hot Air Rises and Heat Sinks: Everything You Kn

About Cooling Electronics is Wrong, ASME, New York.f22g Remsburg, R., 2001,Thermal Design of Electronic Equipment, CRC, Boca

Raton.f23g Cox, C. S., 1987,The Aerobiological Pathway of Microorganisms, Wiley-

Interscience, New York.f24g Clark, R. P., and Cox, R. N., 1973, “The Generation of Aerosols from

Human Body,” inAirborne Transmission and Airborne Infection: Conceand Methods, J. F. Hers and K. C. Winkler, eds., Wiley, New York, pp. 4426.

f25g Burge, H. A., 1995, “Aerobiology of the Indoor Environment,” Occup. M10s1d, pp. 27–40.

f26g Higson, S. P. J., 2003,Analytical Chemistry, Oxford University, Oxford, UKf27g Kline, S. J., 1960, “The Importance of Flow Visualization,”Symposium o

Flow Visualization, ASME, New York, 1, pp. 1–9.f28g Roshko, A., 1995, “Flow Visualization As a Basic Research Tool,” inFlow

Visualization VII, Crowder, J. P. ed., Begell House, New York, pp. 983–9f29g Merzkirch, W., 1987,Flow Visualization, Academic, New York.f30g Paulsen, E., 1882, “Experimentelle Untersuchungen Über Die Strömunge

Luft in Der Nasenhöhle,” Sitzungsber. K. Preuss. Akad. Wiss.,85, p. 348ff.f31g Bojsen-Møller, F., and Fahrenkrug, J., 1971, “Nasal Swell-Bodies and C

Changes in the Air Passage of the Rat and Rabbit Nose,” J. Anat.,110s1d, pp.25–37.

f32g Hornung, D. E., Leopold, D. A., Youngentob, S. L., Sheehe, P. R., GagnM., Thomas, F. D., and Mozell, M. M., 1987, “Airflow Patterns in a HumNasal Model,” Arch. Otolaryngol. Head Neck Surg.,113s2d, pp. 169–172.

f33g Hahn, I., Scherer, P. W., and Mozell, M. M., 1993, “Velocity Profiles Measfor Airflow Through a Large-Scale Model of the Human Nasal CavityAppl. Physiol., 75s5d, pp. 2273–2287.

f34g Keyhani, K., Scherer, P. W., and Mozell, M. M., 1997, “Numerical Simulaof Airflow in the Human Nasal Cavity,” ASME J. Biomech. Eng.,117, pp.429–441.

f35g Brueggemann, J., and Jeckstadt, S., 1938, “Über Den Weg Des LuftsDurch Die Nase Des HundessOn the Direction of the Airstream Through tDog’s Nosed,” Pflugers Archiv, European Journal of Physiology,239, pp.293–303.

f36g Dawes, J. D. K., 1952, “The Course of the Nasal Airstreams,” J. LaryOtol., 66s12d, pp. 583–593.

f37g Ishida, H., Yamanaka, T., Kushida, N., Nakamoto, T., and Moriizumi, T., 2“Study of Real-Time Visualization of Gas/Odor Flow Image Using Gas SeArray,” Sens. Actuators B,65s1–3d, pp. 14–16.

f38g Yamanaka, T., Ishida, H., Nakamoto, T., and Moriizumi, T., 1998, “AnalysGas Sensor Transient Response by Visualizing Instantaneous Gas Contion Using Smoke,” Sens. Actuators, A,69s1d, pp. 77–81.

f39g Saunders, C. J., and Fletcher, B., 1996, “Jet Enhanced Local Exhaust Vtion,” Ann. Occup. Hyg.,37s1d, pp. 15–24.

f40g Teichmann, H., 1959, “Vergleichende Untersuchungen an Der Nase DersComparative Investigations on the Nose of the Fishd,” Z. Morphol. OekolTiere, 43, pp. 171–212.

f41g Grasso, F. W., and Basil, J. A., 2002, “How Lobsters, Crayfishes, andLocate Sources of Odor: Current Perspectives and Future Directions,”Opin. Neurobiol., 12s6d, pp. 721–727.

f42g Atema, J., 1996, “Eddy Chemotaxis and Odor Landscapes: ExploratiNature with Animal Sensors,” Biol. Bull.,191s1d, pp. 129–138.

f43g Webster, D. R., and Weissburg, M. J., 2001, “Chemosensory Guidance Ca Turbulent Chemical Odor Plume,” Limnol. Oceanogr.,46s5d, pp. 1034–1047.

f44g Stuiver, M., 1958, “Biophysics of the Sense of Smell,” PhD thesis, Univeof Groningen, Germany.

f45g Schreck, S., Sullivan, K. J., Ho, C. M., and Chang, H. K., 1993, “CorrelaBetween Flow Resistance and Geometry in a Model of the Human NosAppl. Physiol., 75s4d, pp. 1767–1775.

f46g Settles, G. S., 2001,Schlieren and Shadowgraph Techniques: Visualizingnomena in Transparent Media, 1st Ed., Springer, New York.

f47g Settles, G. S., Kester, D. A., and Dodson-Dreibelbis, L. J., 2003, “The ExAerodynamics of Sniffing,” inSensors and Sensing in Biology and Engining, F. G. Barth et al., eds., Springer, New York, pp. 323–355.

f48g Board on Chemical Sciences and Technology, 2004, “Existing and PoStandoff Explosives Detection Technologies,” US National Research CoReport, National Academies Press, Washington DC.

f49g White, F. M., 2002,Fluid Mechanics, McGraw-Hill, Boston.f50g Batchelor, G. K., 2000,An Introduction to Fluid Mechanics, Cambridge Uni

versity, Cambridge, UK.f51g Chahine, A., Chahine, G. L., Duraiswami, R., Billard, J. Y., and Nieuwsta

T. M., 1995, “Two-Dimensional Potential Flow Instruction and AnalysisPersonal Computers,” ASME J. Fluids Eng.,220, pp. 89–93.

f52g Kimbell, J. S., Gross, E. A., Joyner, D. R., Godo, M. N., and Morgan, K1997, “Application of Computational Fluid Dynamics to Regional Dosimof Inhaled Chemicals in the Upper Respiratory Tract of the Rat,” Tox
Appl. Pharmacol.,121s2d, pp. 253–263.

. BAll

998e in

Fluge

03,atio

n

ode

pp

hatepp.

y of

atus

Proeuro

yng

podwit

-916-

art

.

68.ts

ch.

nta

ing

f FiEx

f FiEx

StrFlu

Sies,”

ch-ent

entaros

., aort

al,t:

ln

rticlesled

Vol.

nsor” J.

ok

sterby

T.,ign

nce,

ection

“Cy-App.

cha-

1996,

and

--314.

s,” in

ncin-

le of

s onrial,”

2001,

,” in

ncin-

ing,”

, andring.

aero-

D.,door

Im-

n-

, R.993,

tdoor

, R.Total-

f53g Kimbell, J. S., Godo, M. N., Gross, E. A., Joyner, D. R., Richardson, Rand Morgan, K. T., 1997, “Computer Simulation of Inspiratory Airflow inRegions of the F344 Rat Nasal Passages,” Toxicol. Appl. Pharmacol.,145s2d,pp. 388–398.

f54g Kepler, G. M., Richardson, R. B., Morgan, K. T., and Kimbell, J. S., 1“Computer Simulation of Inspiratory Nasal Airflow and Inhaled Gas Uptaka Rhesus Monkey,” Toxicol. Appl. Pharmacol.,150s1d, pp. 1–11.

f55g Kimbell, J. S., and Subramaniam, R. P., 2001, “Use of ComputationalDynamics Models for Dosimetry of Inhaled Gases in the Nasal PassaInhalation Toxicol., 13s5d, pp. 325–334.

f56g Koch, D. A., Arnold, S., Hubler, M., and Montavon, P. M., 20“Brachycephalic Syndrome in Dogs,” Compendium on Continuing Educfor the Practicing Veterinarian,25s1d, pp. 48.

f57g Brockmann, J. E., 2001, “Sampling and Transport of Aerosols,” inAerosolMeasurement: Principles, Techniques, and Applications, 2nd Ed., P. A. Baroand K. Willeke, eds., Wiley, New York, pp. 143–195.

f58g Hahn, I., Scherer, P. W., and Mozell, M. M., 1994, “A Mass-Transport Mof Olfaction,” J. Theor. Biol.,167s2d, pp. 115–128.

f59g Anonymous, 2001, “Odors,” inASHRAE Fundamentals Handbook, AmericanSociety of Heating, Ventilation, and Air-Conditioning Engineers, Atlanta,13.1–13.8.

f60g Mozell, M. M., Hornung, D. E., Sheehe, P. R., and Kurtz, D. B., 1986, “WShould be Controlled in Studies of Smell?,” inClinical Measurement of Tasand Smell, H. L. Meiselman and R. S. Rivlin, eds., MacMillan, New York,154–169.

f61g Zuschneid, K., 1973, “The Olfactory Acuity of the DogsDie RiechleistungDes Hundesd,” PhD thesis, School of Veterinary Medicine, Free UniversitBerlin, Berlin, Germany.

f62g Morrison, E. E., 2000, “Canine Detection of Illicit Drugs: Sensory ApparTechnology,” Battelle Report A185183, Columbus, OH.

f63g Johnson, B. N., Mainland, J. D., and Sobel, N., 2003, “Rapid Olfactorycessing Implicates Subcortical Control of an Olfactomotor System,” J. Nphysiol., 90s2d, pp. 1084–1094.

f64g Schreider, J. P., and Raabe, O. G., 1981, “Anatomy of the Nasal-PharAirway of Experimental Animals,” Anat. Rec.,200, pp. 195–205.

f65g Mead, K. S., Koehl, M. A. R., and O’Donnell, M. J., 1999, “StomatoSniffing: the Scaling of Chemosensory Sensillae and Flicking BehaviorBody Size,” J. Exp. Mar. Biol. Ecol.,241s2d, pp. 235–261.

f66g Vogel, S., 1994,Life in Moving Fluids: The Physical Biology of Flow, Prince-ton University, Princeton, NJ.

f67g Olander, L., 2001, “Exhaust Hoods,” inIndustrial Ventilation Design Guidebook, H. D. Goodfellow, and E. Tähti, eds., Elsevier, New York, pp. 814–

f68g Anonymous, 1999, “Industrial Local Exhaust Systems,” inASHRAE Applications Handbook, ASHRAE, Atlanta, pp. 29.1–29.21.

f69g Jackson, C. N., Sherlock, C. N., and Moore, P. O., 1998, “Leak Testing, PTest Techniques Providing Electronic Indications of Leak Locations,” inNon-destructive Testing Handbook, Second Edition, ASNT, Columbus, OH, 1, pp30–33.

f70g List, E. J., 1982, “Mechanics of Turbulent Buoyant Jets and Plumes,” inTur-bulent Buoyant Jets and Plumes, W. Rodi, ed., Pergamon, New York, pp. 1–

f71g Abramovich, G. N., and Schindel, L., 1963,The Theory of Turbulent Je,MIT, Cambridge, MA.

f72g Zhivov, A., 2001, “Air Jets,” inIndustrial Ventilation Design Guidebook, H.D. Goodfellow, and E. Tähti, eds., Elsevier, New York, pp. 446–517.

f73g Glezer, A., and Amitay, M., 2002, “Synthetic Jets,” Annu. Rev. Fluid Me34, pp. 503–529.

f74g Lai, H. Z., Naughton, J. W., and Lindberg, W. R., 2003, “An ExperimeInvestigation of Starting Impinging Jets,” ASME J. Fluids Eng.,125s2d, pp.275–282.

f75g Glauert, M. B., 1956, “The Wall Jet,” J. Fluid Mech.,1s5d, pp. 625–643.f76g Phares, D. J., Smedley, G. T., and Flagan, R. C., 2000, “The Inviscid Imp

ment of a Jet with Arbitrary Velocity Profile,” Phys. Fluids,12s8d, pp. 2046–2055.

f77g Smedley, G. T., Phares, D. J., and Flagan, R. C., 1999, “Entrainment oParticles From Surfaces by Gas Jets Impinging at Normal Incidence,”Fluids, 26s4d, pp. 324–334.

f78g Smedley, G. T., Phares, D. J., and Flagan, R. C., 2001, “Entrainment oParticles From Surfaces by Gas Jets Impinging at Oblique Incidence,”Fluids, 30s2d, pp. 135–142.

f79g Phares, D. J., Smedley, G. T., and Flagan, R. C., 2000, “The Wall ShearProduced by the Normal Impingement of a Jet on a Flat Surface,” J.Mech., 418, pp. 351–375.

f80g Phares, D. J., Smedley, G. T., and Flagan, R. C., 2000, “Effect of Particleand Material Properties on Aerodynamic Re-Suspension From SurfacAerosol Sci., 31s11d, pp. 1335–1353.

f81g Ibrahim, A. H., Dunn, P. F., and Brach, R. M., 2003, “Microparticle Detament From Surfaces Exposed to Turbulent Air Flow: Controlled Experimand Modeling,” J. Aerosol Sci.,34s6d, pp. 765–782.

f82g Ziskind, G., Yarin, L. P., Peles, S., and Gutfinger, C., 2002, “ExperimInvestigation of Particle Removal From Surfaces by Pulsed Air Jets,” AeSci. Technol.,36s5d, pp. 652–659.

f83g Settles, G. S., Thomas, M. A., Ferree, H. C., Jonassen, D. R., Miller, J. DDodson, L. J., 2005, “The Aerodynamics of a Chemical Trace Detection Pfor Screening People,” in preparation.

f84g Marple, V. A., Olson, B. A., and Rubow, K. L., 2001, “Inertial, GravitationCentrifugal, and Thermal Collection Techniques,” inAerosol Measuremen
Principles, Techniques, and Applications, 2nd Ed., P. A. Baron and K. Willeke,

.,

,

ids,”

n

l

.

t

--

eal

h

.

3:

,

l

e-

nep.

nep.

essid

zeJ.

s

lol

ndal

eds., Wiley, New York, pp. 229–260.f85g Baron, P. A., and Willeke, K., 2001, “Gas and Particle Motion,” inAeroso

Measurement: Principles, Techniques, and Applications, 2nd Ed., P. A. Baroand K. Willeke, eds., Wiley, New York, pp. 61–82.

f86g Morawska, L., and Salthammer, T., 2003, “Fundamentals of Indoor Paand Settled Dust,” inIndoor Environment: Airborne Particles and SettDust, L. Morowska and T. Salthammer, eds., Wiley, New York.

f87g Feynman, R. P., 1999, “Diffusion,” inThe Feynman Lectures on Physics,6, Kinetics and Heat, Audio Recording, Basic Books, New York.

f88g Stitzel, S. E., Stein, D. R., and Walt, D. R., 2003, “Enhancing Vapor SeDiscrimination by Mimicking a Canine Nasal Cavity Flow Environment,Am. Chem. Soc.,125s13d, pp. 3684–3685.

f89g Neuhaus, W., 1981, “The Importance of Sniffing in Canine OlfactionsDieBedeutung des Schnüffelns für das Riechen des Hundesd,” Zeitschrift fürSaugetierkunde,46, pp. 301–310.

f90g Vogel, S., 2003,Comparative Biomechanics: Life’s Physical World, PrincetonUniversity, Princeton, NJ.

f91g Anonymous, 2001, “Mass Transfer,” inASHRAE Fundamentals Handbo,ASHRAE, Atlanta, pp. 5.1–5.13.

f92g Goldman, J. A., and Koehl, M. A. R., 2001, “Fluid Dynamic Design of LobOlfactory Organs: High Speed Kinematic Analysis of Antennule FlickingPanulirus Argus,” Chem. Senses,26s4d, pp. 385–398.

f93g Koehl, M. A. R., Koseff, J. R., Crimaldi, J. P., McCay, M. G., Cooper,Wiley, M. B., and Moore, P. A., 2001, “Lobster Sniffing: Antennule Desand Hydrodynamic Filtering of Information in an Odor Plume,” Scie294s5548d, pp. 1948–1951.

f94g Griffy, T. A., 1992, “A Model of Explosive Vapor Concentration II,” inPro-ceedings of the Fourth International Symposium on the Analysis and Detof Explosives, J. Yinon, ed., Kluwer-Academic, New York, pp. 503–511.

f95g Lugt, H. J., 1983,Vortex Flow in Nature and Technology, Wiley, New York.f96g Motchkine, V. S., Krasnobaev, L. Y., and Bunker, S. N., Aug. 21, 2003,

clone Sampling Nozzle for an Ion Mobility Spectrometer,” U.S. PatentUS2003/0155506A1.

f97g Colehour, J. L., and Farquhar, B. W., 1971, “Inlet Vortex,” J. Aircr.,8s1d, pp.39–43.

f98g De Siervi, F., Viguier, H. C., Greitzer, E. M., and Tan, C. S., 1982, “Menisms of Inlet Vortex Formation,” J. Fluid Mech.,124, pp. 173–207.

f99g Chen, P., Soundra-Nayagam, M., Bolton, A. N., and Simpson, H. C.,“Unstable Flow in Centrifugal Fans,” ASME J. Fluids Eng.,118s1d, pp. 128–133.

f100g Stong, C. L., 1971, “Experiments With Wind: A Pendulum AnemometerMiniature Tornadoes,” Sci. Am.,224s10d, pp. 108–112.

f101g Lippmann, M., 2001, “Filters and Filter Holders,” inAir Sampling Instruments for Evaluation of Atmospheric Contaminants, 9th Ed., American Conference of Governmental Industrial Hygienists, Cincinnati, OH, pp. 281–

f102g Hering, S. V., 2001, “Impactors, Cyclones, and Other Particle CollectorAir Sampling Instruments for Evaluation of Atmospheric Contaminants, 9thEd., American Conference of Governmental Industrial Hygienists, Cinati, OH, pp. 315–375.

f103g Johnston, J. M., Morrison, E. E., and Boussom, T. M., 2002, “The RoParticles in Canine Olfactory Detection,” unpublished.

f104g Chang, M. C., Kim, S., and Sioutas, C., 1999, “Experimental StudieParticle Impaction and Bounce: Effects of Substrate Design and MateAtmos. Environ., 33s15d, pp. 2313–2322.

f105g Misra, C., Geller, M. D., Shah, P., Sioutas, C., and Solomon, P. A.,“Development and Evaluation of a Continuous CoarsesPM10-PM2.5d Par-ticle Monitor,” J. Air Waste Manage. Assoc.,51s9d, pp. 1309–1317.

f106g Macher, J. M., and Burge, H. A., 2001, “Sampling Biological AerosolsAir Sampling Instruments for Evaluation of Atmospheric Contaminants, 9thEd., American Conference of Governmental Industrial Hygienists, Cinati, OH, pp. 662–701.

f107g Solomon, W. R., 2003, “How Ill the Wind? Issues in Aeroallergen SamplJ. Allergy Clin. Immunol., 112s1d, pp. 3–8.

f108g Mehta, S. K., Bell-Robinson, D. M., Groves, T. O., Stetzenbach, L. D.Pierson, D. L., 2000, “Evaluation of Portable Air Samplers for MonitoAirborne Culturable Bacteria,” Am. Ind. Hyg. Assoc. J.,61s6d, pp. 850–854

f109g Lacey, J., 1997, “Guest Editorial: Sampling and Rapid Assay of Biosols,” J. Aerosol Sci.,28s3d, pp. 345–348.

f110g Griffiths, W. D., Stewart, I. W., Futter, S. J., Upton, S. L., and Mark,1997, “Development of Sampling Methods for the Assessment of InBioaerosols,” J. Aerosol Sci.,28s3d, pp. 437–457.

f111g Heinsohn, R. J., and Cimbala, J. M., 2002,Indoor Air Quality Engineering,Marcel Dekker, New York.

f112g Wallace, L. A., 2001, “Human Exposure to Volatile Organic Pollutants:plications for Indoor Air Studies,” Annu. Rev. Energy Envir.,26, pp. 269–301.

f113g Anonymous, 2001, “Indoor Environmental Health,” inASHRAE Fundametals Handbook, ASHRAE, Atlanta, GA, pp. 9.1–9.20.

f114g Thomas, K. W., Pellizzari, E. D., Clayton, C. A., Whitaker, D. A., ShoresC., Spengler, J., Ozkaynak, H., Froelich, S. E., and Wallace, L. A., 1“Particle Total Exposure Assessment MethodologysPTEAMd 1990 Study—Method Performance and Data Quality for Personal, Indoor, and OuMonitoring,” J. Expo Anal Environ. Epidemiol.,3s2d, pp. 203–226.

f115g Clayton, C. A., Perritt, R. L., Pellizzari, E. D., Thomas, K. W., WhitmoreW., Wallace, L. A., Ozkaynak, H., and Spengler, J. D., 1993, “ParticleExposure Assessment MethodologysPTEAMd Study-Distributions of Aero
sol and Elemental Concentrations in Personal, Indoor, and Outdoor Air
MARCH 2005, Vol. 127 / 213

Epi

nkinFro

ron

ancsme

969

t

tionns.

ch-iron

nalsnive

orn. F

odand32.rne

man2,,

lingffica

elop

, anhod

A.Pa

t thHyg

Arling1–

Evi-

U.S

, Paeliu-

n-

-H/

lac

Ve-

. S.Bio-nal.

Kingnsi

dleyD. JTro

craftphys.

g Ef-ted,

ro-

f Mi-

ter-onalttp://

nalttp://

ativeorne

De-and

.the

and

Shown-

ls

ffects

. A.,s of

Near0-5,

-

,

and-olo-a-898.nd

nt. J.

s,” inions,

ts,”ew

sory

oralomol.,

es,”

. S.,iga-

ree-

rrish,ess inobe,”

en-

ne-rack-xp.

How

Samples in a Southern California Community,” J. Expo Anal Environ.demiol., 3s2d, pp. 227–250.

f116g Ozkaynak, H., Xue, J., Spengler, J., Wallace, L., Pellizzari, E., and JeP., 1996, “Personal Exposure to Airborne Particles and Metals: Resultsthe Particle Team Study in Riverside, California,” J. Expo Anal EnviEpidemiol., 6s1d, pp. 57–78.

f117g Rodes, C. E., Kamens, R. M., and Wiener, R. W., 1991, “The Significand Characteristics of the Personal Activity Cloud on Exposure AssesMeasurements for Indoor Contaminants,” Indoor Air,2, pp. 123–145.

f118g Wallace, L. A., 1993, “The TEAM Studies,” EPA Journal,19s4d.f119g Lewis, H. E., Foster, A. R., Mullan, B. J., Cox, R. N., and Clark, R. P., 1

“Aerodynamics of the Human Microenvironment,” Lancet,322s7609d, pp.1273–1277.

f120g Clark, R. P., and Edholm, O. G., 1985,Man and His Thermal Environmen,E. Arnold, London.

f121g Homma, H., and Yakiyama, M., 1991, “Examination of Free ConvecAround Occupant’s Body Caused by Its Metabolic Heat,” ASHRAE Tra94 Part 1s3118d, pp. 104–124.

f122g Clark, R. P., and Cox, R. N., 1974, “An Application of Aeronautical Teniques to Physiology. 2. Particle Transport within the Human Microenvment,” Med. Biol. Eng.,12s3d, pp. 275–279.

f123g Gowadia, H. A., 2000, “The Natural Sampling of Airborne Trace SigAssociated with the Human Body,” PhD thesis, Pennsylvania State Usity, University Park, PA.

f124g Gowadia, H. A., and Settles, G. S., 2001, “The Natural Sampling of AirbTrace Signals From Explosives Concealed Upon the Human Body,” Jrensic Sci.,46s6d, pp. 1324–1331.

f125g Syrotuck, W. G., 1972,Scent and the Scenting Dog, Arner, Rome, NY.f126g May, K. R., and Pomeroy, N. P., 1973, “Bacterial Dispersion from the B

Surface,” inAirborne Transmission and Airborne Infection: ConceptsMethods, J. F. Hers and K. C. Winkler, eds., Wiley, New York, pp. 426–4

f127g Clark, R. P., and Shirley, S. G., 1973, “Identification of Skin in AirboParticulate Matter,” NaturesLondond, 246s5427d, pp. 39–40.

f128g Clark, R. P., 1974, “Skin Scales Among Airborne Particles,” J. Hyg.sLondd,72s1d, pp. 47–51.

f129g Sykes, B., 2001,The Seven Daughters of Eve, Norton, New York.f130g Murakami, S., 2002, “CFD Study on the Micro-Climate Around the Hu

Body with Inhalation and Exhalation,” inProceedings of ROOMVENT 2008th International Conference on Air Distribution in Rooms, Copenhagen8–11, Technical University of Denmark.

f131g Renstrom, A., Karlsson, A. S., and Tovey, E., 2002, “Nasal Air Sampused for the Assessment of Occupational Allergen Exposure and the Eof Respiratory Protection,” Clin. Exp. Allergy,32s12d, pp. 1769–1775.

f132g Misra, C., Singh, M., Shen, S., Sioutas, C., and Hall, P. A., 2002, “Devment and Evaluation of a Personal Cascade Impactor SamplersPCISd,” J.Aerosol Sci., 33s7d, pp. 1027–1047.

f133g Hauck, B. C., Grinshpun, S. A., Reponen, A., Reponen, T., Willeke, K.Bornschein, R. L., 1997, “Field Testing of New Aerosol Sampling MetWith a Porous Curved Surface As Inlet,” Am. Ind. Hyg. Assoc. J.,58s10d,pp. 713–719.

f134g Aizenberg, V., Choe, K., Grinshpun, S. A., Willeke, K., and Baron, P.2001, “Evaluation of Personal Aerosol Samples Challenged with Largeticles,” J. Aerosol Sci.,32s6d, pp. 779–793.

f135g Guffey, S. E., Flanagan, M. E., and van Belle, G., 2001, “Air Sampling aChest and Ear As Representative of the Breathing Zone,” Am. Ind.Assoc. J.,62s4d, pp. 416–427.

f136g Fletcher, B., and Johnson, A. E., 1989, “Comparison of Personal andConcentration of Measurements, and the Use of a Mannequin in SampProceedings Ventilation ’88, Vincent, J. H., ed., BOHS, London, pp. 16167.

f137g Fetterolf, D. D., and Clark, T. D., 1993, “Detection of Trace Explosivedence by Ion Mobility Spectrometry,” J. Forensic Sci.,38s1d, pp. 28–39.

f138g Baret, G., and Gevaud, R., Aug. 30, 1994, “Helium Leak Detector,”Patent 5,341,671.

f139g Jackson, C. N., Sherlock, C. N., and Moore, P. O., 1998, “Leak Testing5: Sampling Detector Probe Testing of Systems Pressurized with HTracer Gas,” inNondestructive Testing Handbook, 2nd Ed., ASNT, Columbus, OH, 1, pp. 478–489.

f140g DeSorbo, M. A., 2001, “Fingerprinting Particles ‘Automatically’,” Clearooms, 15s10d, pp. 1–54.

f141g Anonymous, 2004, “Portable Gas Sample Extraction Probe PSP4000T,” M&C Products Analysertechnik Gmbh, Ratingen, Germany.

f142g Grasmeyer, J. M., and Keennon, M. T., 2001, “Development of the BWidow Micro Air Vehicle,” AIAA Paper No. 2001-0127.

f143g Mueller, T. J., and DeLaurier, J. D., 2003, “Aerodynamics of Smallhicles,” Annu. Rev. Fluid Mech.,35, pp. 89–111.

f144g Anderson, G. P., King, K. D., Cuttino, D. S., Whelan, J. P., Ligler, FMacKrell, J. F., Bovais, C. S., Indyke, D. K., and Foch, R. J., 1999, “logical Agent Detection With the Use of an Airborne Biosensor,” Field AChem. Tech., 3s4–5d, pp. 307–314.

f145g Ligler, F. S., Anderson, G. P., Davidson, P. T., Foch, R. J., Ives, J. T.,K. D., Page, G., Stenger, D. A., and Whelan, J. P., 1998, “Remote SeUsing an Airborne Biosensor,” Environ. Sci. Technol.,32, pp. 2461–2466.

f146g Brune, W. H., Faloona, I. C., Tan, D., Weinheimer, A. J., Campos, T., RiB. A., Vay, S. A., Collins, J. E., Sachse, G. W., Jaegle, L., and Jacob,1998, “Airborne in-Situ OH and HO2 Observations in the Cloud-Free
posphere and Lower Stratosphere During SUCCESS,” Geophys. Res. Le
214 / Vol. 127, MARCH 2005

-

s,m

.

ent

,

,

-

r-

eo-

y

cy

-

d

,r-

e.

ea,”

.

rtm

C/

k

,

,ng

,.,

-

25s10d, pp. 1701–1704.f147g Porter, J. N., Clarke, A. D., Ferry, G., and Pueschel, R. F., 1992, “Air

Studies of Size-Dependent Aerosol Sampling Through Inlets,” J. GeoRes.,fAtmos.g, 97sD4d, pp. 3815–3824.

f148g Sheridan, P. J., and Norton, R. B., 1998, “Determination of the Passinficiency for Aerosol Chemical Species Through a Typical Aircraft-MounDiffuser-Type Aerosol Inlet System,” J. Geophys. Res.,fAtmos.g, 103sD7d,pp. 8215–8225.

f149g Reed, R. D., 1978, “High-Flying Mini-Sniffer RPV—Mars Bound,” Astnaut. Aeronaut.,16s6d, pp. 26–39.

f150g Spedding, G. R., and Lissaman, P. B. S., 1998, “Technical Aspects ocroscale Flight Systems,” Journal of Avian Biology,29s4d, pp. 458–468.

f151g Singh, H., Jacob, D., Pfister, L., and Crawford, J., 2002, “INTEX-NA: Incontinental Chemical Transport Experiment—North America,” NatiAeronautics and Space Administration While Paper, hcloud1.arc.nasa.gov/intex-na/overview/whiteIpaper.pdf.

f152g Kiel, G., 1935, “Total-Head Meter with Small Sensitivity to Yaw,” NatioAdvisory Committee for Aeronautics, Technical Memorandum 775, hnaca.larc.nasa.gov/reports/1935/naca-tm-775/

f153g Cain, S. A., Ram, M., and Woodward, S., 1998, “Qualitative and QuantitWind Tunnel Measurements of the Airflow Through a Shrouded AirbAerosol Sampling Probe,” J. Aerosol Sci.,29s9d, pp. 1157–1169.

f154g Soderman, P. T., Hazen, N. L., and Brune, W. H., 1991, “Aerodynamicsign of Gas and Aerosol Samplers for Aircraft,” National AeronauticsSpace Administration Technical Memorandum 103854.

f155g Sutton, O. G., 1953,Micrometeorology, McGraw-Hill, New York.f156g Arya, P. S., 2001,Introduction to Micrometeorology, Academic, New Yorkf157g DeBlieu, J., 1998,Wind: How Air Flow Has Shaped Life, Myth, and

Land, Houghton Mifflin, Boston, MA.f158g Shivik, J. A., 2002, “Odor-Adsorptive Clothing, Environmental Factors,

Search-Dog Ability,” Wildlife Soc. Bull., 30s3d, pp. 721–727.f159g Schouest, L. P., and Miller, T. A., 1994, “Automated Pheromone Traps

Male Pink-BollwormsLepidoptera, Gelechiidaed Mating Response Is Depedent on Weather Conditions,” J. Econ Entomol.,87s4d, pp. 965–974.

f160g Anonymous, 2001, “Airflow Around Buildings,” inASHRAE FundamentaHandbook, ASHRAE, Atlanta, GA, pp. 16.1–16.12.

f161g Vander Wall, S. B., 1998, “Foraging Success of Granivorous Rodents: Eof Variation in Seed and Soil Water on Olfaction,” Ecology,79s1d, pp. 233–241.

f162g Jenkins, T. F., Walsh, M. E., Miyares, P. H., Kopczinski, J. A., Ranney, TGeorge, V., Pennington, J. C., and Berry, T. E., 2000, “AnalysiExplosives-Related Chemical Signatures in Soil Samples CollectedBuried Land Mines,” U.S. Army Corps of Engineers Report ERDC TR-0Hanover, NH.

f163g Turner, J. S., 1973,Buoyancy Effects in Fluids, Cambridge University, Cambridge, UK.

f164g List, E. J., 1982, “Turbulent Jets and Plumes,” Annu. Rev. Fluid Mech.14,pp. 189–212.

f165g Scorer, R. S., 1958,Natural Aerodynamics, Pergamon, New York.f166g Settles, G. S., and Kester, D. A., 2001, “Aerodynamic Sampling for L

mine Trace Detection,” inProc. SPIE—Detection and Remediation Techngies for Mines and Minelike Targets, VI, Dubey, A. C., et al., eds., Interntional Society for Optical Engineering, Bellingham, WA, 4394, pp. 890–

f167g Mollendorf, J. C., Arif, H., and Ajiniran, E. B., 1996, “Developing Flow aTransport Above a Suddenly Heated Horizontal Surface in Water,” IHeat Mass Transfer,27s2d, pp. 273–289.

f168g Carde, R. T., 1996, “Odour Plumes and Odour-Mediated Flight in InsectCiba Foundation Symposium on Olfaction in Mosquito-Host InteractAssociated Scientific, Amsterdam, pp. 54–70.

f169g Ishida, H., and Moriizumi, T., 2003, “Machine Olfaction for Mobile Roboin Handbook of Machine Olfaction, T. C. Pearce et al., eds., Wiley, NYork, pp. 399–417.

f170g Weissburg, M. J., 2000, “The Fluid Dynamical Context of ChemosenBehavior,” Biol. Bull., 198s2d, pp. 188–202.

f171g Murlis, J., Willis, M. A., and Carde, R. T., 2000, “Spatial and TempStructures of Pheromone Plumes in Fields and Forests,” Phys. Ent25s3d, pp. 211–222.

f172g Vickers, N. J., 2000, “Mechanisms of Animal Navigation in Odor PlumBiol. Bull., 198s2d, pp. 203–212.

f173g Finelli, C. M., Pentcheff, N. D., Zimmer-Faust, R. K., and Wethey, D1999, “Odor Transport in Turbulent Flows: Constraints on Animal Navtion,” Limnol. Oceanogr.,44s4d, pp. 1056–1071.

f174g Moore, P. A., and Atema, J., 1991, “Spatial Information in the ThDimensional Fine Structure of an Aquatic Odor Plume,” Biol. Bull.,181s12d,pp. 408–418.

f175g Moore, P. A., Zimmer-Faust, R. K., Bement, S. L., Weissburg, M. J., PaJ. M., and Gerhardt, G. A., 1992, “Measurement of Microscale Patchina Turbulent Aquatic Odor Plume Using a Semiconductor-Based MicroprBiol. Bull., 183s8d, pp. 138–142.

f176g Koehl, M. A. R., 1996, “Small-Scale Fluid Dynamics of Olfactory Antnae,” Marine Freshwater Behav. Physiol.,27s2–3d, pp. 127–141.

f177g Mead, K. S., Wiley, M. B., Koehl, M. A. R., and Koseff, J. R., 2003, “FiScale Patterns of Odor Encounter by the Antennules of Mantis Shrimp Ting Turbulent Plumes in Wave-Affected and Unidirectional Flow,” J. EBiol., 206s1d, pp. 181–193.

f178g Murlis, J., Elkington, J. S., and Carde, R. T., 1992, “Odor Plumes and
tt., Insects Use Them,” Annu. Rev. Entomol.,37, pp. 505–532.

vingEco

lumiol.

g oobil

001iol.

ens

tedone

1,

Odo

dap

s inrity,t Rrial796a Nare

urc

idplied

. L.isoull.

ideind.

arin

ose

23

ture

tion

erglianNa

ds

turey

and

errotion

cha

.lian

ysi

s i

ure

nsc

ay

994,atory

. E.,ine-

ech-.

dge,s for

dge,nine

ogs

n in

,

rks byn-

d ofhsych.,

r, D.

on ofol.

rials

k!,”

s ofZh.,

mell

man

gol.,

ol.,

cine,

tory,

n We

e Be-

deleor.

andn-

. E.,

R. L.,arate

990,ar-

f179g Weissburg, M. J., and Zimmer-Faust, R. K., 1993, “Life and Death in MoFluids: Hydrodynamic Effects on Chemosensory-Mediated Predation,”ogy, 74, pp. 1428–1443.

f180g Geier, M., Bosch, O. J., and Boeckh, J., 1999, “Influence of Odour PStructure on Upwind Flight of Mosquitoes Towards Hosts,” J. Exp. B202s12d, pp. 1639–1648.

f181g Ishida, H., Nakamoto, T., and Moriizumi, T., 1998, “Remote SensinGas/Odor Source Location and Concentration Distribution Using MSystem,” Sens. Actuators B,49s1–2d, pp. 52–57.

f182g Ishida, H., Nakamoto, T., Moriizumi, T., Kikas, T., and Janata, J., 2“Plume-Tracking Robots: A New Application of Chemical Sensors,” BBull., 200s2d, pp. 222–226.

f183g Deveza, R., Thiel, D., Russell, A., and Mackay-Sim, A., 1994, “Odor Sing for Robot Guidance,” Int. J. Robot. Res.,13, pp. 232–239.

f184g Li, W., Farrell, J. A., and Carde, R. T., 2001, “Tracking of Fluid-AdvecOdor Plumes: Strategies Inspired by Insect Orientation to PheromAdaptive Behavior,9s3–4d, pp. 143–170.

f185g Dusenbery, D. B., 2003, “Mechanisms for Gradient Following,” inSensorsand Sensing in Biology and Engineering, Springer-Verlag, Vienna, Chap. 2pp. 291–303.

f186g Grasso, F., Consi, T., Mountain, D., and Atema, J., 1996, “LocatingSources in Turbulence with a Lobster Inspired Robot,” inFrom Animals toAnimats: Proceedings of the International Conference on Simulation Ative Behavior, P. Maes et al., eds., MIT, Cambridge, MA, pp. 104–112.

f187g Anonymous, 2003, “Following Toxic Clouds: Science and AssumptionPlume Modeling” Hearing Before Subcommittee on National SecuEmerging Threats & International Relations, Committee on Governmenform, U.S. House of Representatives, 108th Congress, 1st Session, Se108-91, US GPO Report 1016-A, http://purl.access.gpo.gov/GPO/LPS4

f188g Sharan, M., and Gopalakrishnan, S. G., 1997, “Bhopal Gas Accident:merical Simulation of the Gas Dispersion Event,” Environ. Model. Softw12s2–3d, pp. 135–141.

f189g Gudiksen, P. H., Harvey, T. F., and Lange, R., 1989, “Chernobyl SoTerm, Atmospheric Dispersion, and Dose Estimation,” Health Phys.,57s5d,pp. 697–706.

f190g List, E. J., 1996, “Environmental Fluid Mechanics,”Research Trends in FluDynamics: Report from US National Commission on Theoretical & ApMechanics, J. L. Lumley ed., Springer, New York, pp. 148–166.

f191g Sullivan, T. J., Ellis, J. S., Foster, C. S., Foster, K. T., Baskett, RNasstrom, J. S., and Schalk, W. W., 1993, “Atmospheric Release AdvCapability-Real-Time Modeling of Airborne Hazardous Materials,” BAm. Meteorol. Soc.,74s12d, pp. 2343–2361.

f192g Cowan, I. R., Castro, I. P., and Robins, A. G., 1997, “Numerical Consations for Simulations of Flow and Dispersion Around Buildings,” J. WEng. Ind. Aerodyn.,67–68, pp. 535–545.

f193g Boris, J., 2002, “The Threat of Chemical and Biological Terrorism: Prepa Response,” Comput. Sci. Eng.,4s2d, pp. 22–32.

f194g Negus, V. E., 1958,The Comparative Anatomy and Physiology of the Nand Paranasal Sinuses, Livingstone, London.

f195g Bang, B. G., 1971, “Functional Anatomy of the Olfactory System inOrders of Birds,” Acta Anat. Suppl.sBaseld, 79, pp. 1–76.

f196g Stacey, M. T., Mead, K. S., and Koehl, M. A. R., 2002, “Molecule Capby Olfactory Antennules: Mantis Shrimp,” J. Math. Biol.,44s1d, pp. 1–30.

f197g U.S. Department of Agriculture, 2004, “Integrated Taxonomic InformaSystem On-Line Database,” http://www.itis.usda.gov.

f198g Arnason, U., Adegoke, J. A., Bodin, K., Born, E. W., Esa, Y. B., GullbA., Nilsson, M., Short, R. V., Xu, X. F., and Janke, A., 2002, “MammaMitogenomic Relationships and the Root of the Eutherian Tree,” Proc.Acad. Sci. U.S.A.,99s12d, pp. 8151–8156.

f199g Carroll, R. L., 1997,Patterns and Processes of Vertebrate Evolution, Cam-bridge University, Cambridge, England.

f200g Wayne, R. K., 1993, “Molecular Evolution of the Dog Family,” TrenGenet., 9s6d, pp. 218–224.

f201g International Committee on Veterinary Gross Anatomical Nomencla1972, Nomina Anatomica Veterinaria, World Association of VeterinarAnatomists, New York.

f202g Wayne, R. K., Leonard, J. A., and Cooper, A., 1999, “Full of SoundFury: The Recent History of Ancient DNA,” Annu. Rev. Ecol. Syst.,30, pp.457–477.

f203g Vogel, S., 1994, “Nature’s Pumps,” Am. Sci.,82s5d, pp. 464–471.f204g Benyus, J. M., 1997,Biomimicry, William Morrow & Co., New York.f205g Moniz, E. J., and Baldeswieler, J. D., 2003, “Approaches to Combat T

ism sACTd: Opportunities for Basic Research,” National Science Foundahttp://www.mitre.org/public/act/10I22Ifinal.pdf

f206g Negus, V. E., 1960, “Further Observations on the Air-Conditioning Menism of the Nose,” Ann. R. Coll. Surg. Engl.,27, pp. 171–204.

f207g Moulton, D. G., 1967, “Olfaction in Mammals,” Am. Zool.,7, pp. 421–429f208g Hillenius, W. J., 1992, “The Evolution of Nasal Turbinates and Mamma

Endothermy,” Paleobiology,18, pp. 17–29.f209g Geist, H. R., 2000, “Nasal Respiratory Turbinate Function in Birds,” Ph

ological and Biochemical Zoology,73s5d, pp. 581–589.f210g Becker, R. F., and King, J. E., 1957, “Delineation of the Nasal Airstream

the Living Dog,” Arch. Otolaryngol.,65s5d, pp. 428–436.f211g Neuhaus, W., 1953, “Über die Riechschärfe des Hundes für Fettsä

Zeitschrift für Vergleichende Physiologie, 35, pp. 527–552.f212g Neuhaus, W., 1955, “Die Unterscheidung von Luftquantitaeten bei Me

und Hund nach Versuchen Mit Buttersäure,” Zeitschrift für Vergleichend


l-

e,

fe

,

-

,”

r

-

e-No.7u-,

e

,ry

r-

g

,

tl.

,

r-,

-

-

n

n,”

h

Physiologie, 37, pp. 234–252.f213g McCartney, W., 1968,Olfaction and Odours: An Osphrésiological Ess,

Springer-Verlag, Berlin.f214g Johnston, J. M., Myers, L. J., Waggoner, L. P., and Williams, M., 1

“Determination of Canine Olfactory Thesholds using Operant LaborMethods,”Proceedings Substance Detection Systems, Proc. SPIE,2092, pp.238–243.

f215g Johnston, J. M., Williams, M., Waggoner, L. P., Edge, C. C., Dugan, Rand Hallowell, S. F., 1998, “Canine Detection Odor Signatures for MRelated Explosives,” inProceedings of SPIE, Detection Remediation Tnology for Mines and Minelike Targets III, Proc. SPIE,3392, Part 1, pp490–501.

f216g Williams, M., Johnston, J. M., Cicoria, M., Paletz, E., Waggoner, L. P., EC. C., and Hallowell, S. F., 1998, “Canine Detection Odor SignatureExplosives,” Proc. SPIE,3575, pp. 291–301.

f217g Williams, M., Johnston, J. M., Cicoria, M., Paletz, E., Waggoner, L. P., EC. C., and Hallowell, S. F., 1998, “Effects of Extraneous Odors on CaDetection,” Proc. SPIE,3575, pp. 355–362.

f218g Thesen, A., Steen, J. B., and Doving, K. B., 1993, “Behaviour of DDuring Olfactory Tracking,” J. Exp. Biol.,180, pp. 247–251.

f219g Steen, J. B., Mohus, I., Kvesetberg, T., and Walloe, L., 1996, “OlfactioBird Dogs During Hunting,” Acta Physiol. Scand.,157s1d, pp. 115–119.

f220g Evans, H. E., 1993,Miller’s Anatomy of the Dog, Saunders, Philadelphia.f221g Anderson, W., Anderson, B. G., and Smith, B. J., 1994,Atlas of Canine

Anatomy, Lea & Febiger, Philadelphia.f222g Thomas, E. M., 1993,The Hidden Life of Dogs, Houghton Mifflin, Boston

MA.f223g Johnston, R. E., and Bhorade, A., 1998, “Perception of Scent Over-Ma

Golden HamsterssMesocricetus Auratusd: Novel Mechanisms for Determiing Which Individual’s Mark is on Top,” J. Comp., Psych.,112s3d, pp. 230–243.

f224g Ferkin, M. H., Dunsavage, J., and Johnston, R. E., 1999, “What KinInformation do Meadow Voles,Microtus Pennsylvanicus, Use to DistinguisBetween the Top and Bottom Scent of an Over-Mark?,” J. Comp., P113s1d, pp. 43–51.

f225g Achter, E. K., Hobbs, J. R., Fine, D. H., Dionne, B. C., and RounbehleP., 1986, “Vapor Pressure of Explosives,” J. Energ. Mater.,4s1–4d, pp. 447–472.

f226g Johnson, W. E., and O’Brien, S. J., 1997, “Phylogenetic Reconstructithe Felidae Using 16S RRNA and NADH-5 Mitochondrial Genes,” J. MEvol., 44sSuppl 1d, pp. S98–S116.

f227g Clifford, A. B., and Witmer, L. M., 2004, “Case Studies in Novel NaAnatomy: 2. The Enigmatic Nose of MoosesArtiodactyla: Cervidae: AlceAlcesd,” J. Zool., 262, pp. 339–360.

f228g Shoshani, J., 1997, “It’s a Nose! It’s a Hand! Its an Elephants’ TrunNatural History, 106s10d, pp. 37–43.

f229g Mariappa, D., 1986,Anatomy and Histology of the Indian Elephant, IndiraPublishing House, Oak Park, MI.

f230g Glebovskii, V. D., and Marevskaya, A. P., 1968, “Participation of Musclethe Nostrils in Olfactory Analysis and Respiration in Rabbits,” Fiziol.54s11d, pp. 1278–1286.

f231g Le Guerer, A., 1992,Scent: The Essential and Mysterious Powers of S,Turtle Bay Books/Random House, New York.

f232g Stoddart, D. M., 1990,The Scented Ape: The Biology and Culture of HuOdour, Cambridge University, Cambridge, England.

f233g Proetz, A. W., 1941,Essays on the Applied Physiology of the Nose, AnnalsPublishing Co., St. Louis, Mo.

f234g Eccles, R., 2000, “Nasal Airflow in Health and Disease,” Acta Oto-Laryn120s5d, pp. 580–595.

f235g Cole, P., 2000, “Biophysics of Nasal Airflow: A Review,” Am. J. Rhin14s4d, pp. 245–249.

f236g Lin, S. J., and Danahey, D. G., 2002, “Nasal Aerodynamics,” eMedi3s4d.

f237g Zimmer, C., 2002, “The Rise and Fall of the Nasal Empire,” Natural His111s5d, pp. 32–35.

f238g Shepherd, G. M., 2004, “The Human Sense of Smell: Are We Better ThaThink?,” PLoS Biology, 2s5d, pp. 572–575.

f239g Haselton, F. R., and Sperandio, P. G. N., 1988, “Convective Exchangtween the Nose and the Atmosphere,” J. Appl. Physiol.,64s6d, pp. 2575–2581.

f240g Keyhani, K., Scherer, P. W., and Mozell, M. M., 1997, “A Numerical Moof Nasal Odorant Transport for the Analysis of Human Olfaction,” J. ThBiol., 186s3d, pp. 279–301.

f241g Simmen, D., Scherrer, J. L., Moe, K., and Heinz, B., 1999, “A DynamicDirect Visualization Model for the Study of Nasal Airflow,” Arch. Otolarygol. Head Neck Surg.,125s9d, pp. 1015–1021.

f242g Sobel, N., Khan, R. M., Saltman, A., Sullivan, E. V., and Gabrieli, J. D1999, “Olfaction-The World Smells Different to Each Nostril,” NaturesLon-dond, 402s6757d, p. 35.

f243g Sobel, N., Prabhakaran, V., Desmond, J. E., Glover, G. H., Goode,Sullivan, E. V., and Gabrieli, J. D. E., 1998, “Sniffing and Smelling: SepSubsystems in the Human Olfactory Cortex,” NaturesLondond, 392s6673d,pp. 282–286.

f244g Cheng, Y. S., Hansen, G. K., Su, Y. F., Yeh, H. C., and Morgan, K. T., 1“Deposition of Ultrafine Aerosols in Rat Nasal Molds,” Toxicol. Appl. Phmacol., 106, pp. 222–233.

ef245g Kelly, J. T., and Asgharian, B., 2003, “Nasal Molds as Predictors of Fine and

MARCH 2005, Vol. 127 / 215

col.

and

is-Pasnce

rona

thesio

987dor

,

: Th. Sc

ni-

ien

s oNav

ls a

ates

andture

nuri

ber:

imeter,”luidu

ayoth

ingion

L. JquitBlo

s: A

., dn aAn

ourAp

f Mo

os

ind

ppl.

ell-

rien-

hol-fer a

nstru-y,

emi-New

nal.

ock-luo-hemi-rans.

P.,De-

l Ol-

and

Land-E–

ion ind2.

uide-ge,

r of

Ion

.,ystem

on

tp://

por

P.,atent

hy:

andn-E,

nce,

lenge

992,

llec-ami-ists,

-, pp.

race

vel-d Its

Coarse Particle Deposition in Rat Nasal Airways,” Inhalation Toxi15s9d, pp. 859–875.

f246g Kelly, J. T., Kimbell, J. S., and Asgharian, B., 2001, “Deposition of FineCoarse Aerosols in a Rat Nasal Mold,” Inhalation Toxicol.,13s7d, pp. 577–588.

f247g Morgan, K. T., Kimbell, J. S., Monticello, T. M., Patra, A. L., and Flechman, A., 1991, “Studies of Inspiratory Air-Flow Patterns in the Nasalsages of the F344 Rat and Rhesus-Monkey Using Nasal Molds-RelevaFormaldehyde Toxicity,” Toxicol. Appl. Pharmacol.,110s2d, pp. 223–240.

f248g Kauer, J. S., 1968, “A Study of the Structure and Function of the Vomesal sJacobson’sd Organ,” MA thesis, Clark University, Worcester, MA.

f249g Welker, W. I., 1964, “Analysis of Sniffing of the Albino Rat,” Behaviour,22,pp. 223–244.

f250g Wilson, D. A., and Sullivan, R. M., 1999, “Respiratory Airflow Pattern atRat’s Snout and an Hypothesis Regarding its Role in Olfaction,” PhyBehav., 66s1d, pp. 41–44.

f251g Youngentob, S. L., Mozell, M. M., Sheehe, P. R., and Hornung, D. E., 1“A Quantitative Analysis of Sniffing Strategies in Rats Performing ODetection Tasks,” Physiol. Behav.,41s1d, pp. 59–69.

f252g Kepecs, A., 2004sunpublishedd.f253g Kleerekoper, H., 1969,Olfaction in Fishes, Indiana University, Bloomington

IN.f254g Pohlmann, K., Grasso, F. W., and Breithaupt, T., 2001, “Tracking Wakes

Nocturnal Predatory Strategy of Piscivorous Catfish,” Proc. Natl. AcadU.S.A., 98s13d, pp. 7371–7374.

f255g Wallraff, H. G., 2003, “Olfactory Navigation by Birds,” Journal für Orthologie, 144s1d, pp. 1–32.

f256g Wallraff, H. G., 2000, “Simulated Navigation Based on Observed Gradof Atmospheric Trace GasessModels on Pigeon Homing, Part 3d,” J. Theor.Biol., 205s1d, pp. 133–145.

f257g Wallraff, H. G., and Andreae, M. O., 2000, “Spatial Gradients in RatioAtmospheric Trace Gases: A Study Stimulated by Experiments on Birdgation,” Tellus, Ser. B,52s4d, pp. 1138–1157.

f258g Bang, B. G., 1966, “The Olfactory Apparatus of Tubenosed BirdssProcel-lariiformesd,” Acta Anat. sBaseld, 65, pp. 391–415.

f259g Pennycuick, C. J., 2002, “Gust Soaring as a Basis for the Flight of PetreAlbatrossessProcellariiformesd,” Avian Science,2s1d, pp. 1–12.

f260g Witmer, L. M., 2001, “A Nose for All Seasons,” Natural History,110, pp.64–71.

f261g Witmer, L. M., 2001, “Nostril Position in Dinosaurs and Other Vertebrand its Significance for Nasal Function,” Science,293, pp. 850–853.

f262g Cruickshank, A. R. I., Small, P., and Taylor, M., 1991, “Dorsal NostrilsHydrodynamically Driven Underwater Olfaction in Plesiosaurs,” NasLondond, 352, pp. 62–64.

f263g Goldman, J. A., and Patek, S. N., 2002, “Two Sniffing Strategies in PaliLobsters,” J. Exp. Biol.,205s24d, pp. 3891–3902.

f264g Koehl, M. A. R., 2001, “Transitions in Function at Low Reynolds NumHair-Bearing Animal Appendages,” Math. Methods Appl. Sci.,24s17-18d,pp. 1523–1532.

f265g Hansson, B. S., 1999,Insect Olfaction, Springer-Verlag, Berlin.f266g Humphrey, J. A. C., and Haj-Hariri, H., 2002, “Detection and Real T

Processing of Odor Plume Information by Arthropods in Air and WaProceedings of the AFI-2002 Mini-Symposium on Advanced FInformation—Fusion of EFD and CFD, Institute of Fluid Science, TohokUniversity, Tokyo, pp. 47–67.

f267g Schneider, R. W. S., Lanzen, J., and Moore, P. A., 1998, “Boundary-LEffect on Chemical Signal Movement Near the Antennae of the Sphinx MManduca Sexta: Temporal Filters for Olfaction,” J. Comp. Physiol.fAg,182s3d, pp. 287–298.

f268g Loudon, C., and Koehl, M. A. R., 2001, “Sniffing by a Silkworm Moth: WFanning Enhances Air Penetration Through and Pheromone InterceptAntennae,” J. Exp. Biol.,203s19d, pp. 2977–2990.

f269g Spielman, A., and D’Antonio, M., 2001,Mosquito, Hyperion, New York.f270g Wright, R. H., 1975, “Why Mosquito Repellents Repel,” Sci. Am.,233s7d,

pp. 104–111.f271g Fox, A. N., Pitts, R. J., Robertson, H. M., Carlson, J. R., and Zwiebel,

2001, “Candidate Odorant Receptors from the Malaria Vector MosAnopheles Gambiae and Evidence of Down-Regulation in Response toFeeding,” Proc. Natl. Acad. Sci. U.S.A.,98s25d, pp. 14693–14697.

f272g Takken, W., 1991, “The Role of Olfaction in Host-Seeking of MosquitoeReview,” Sci. App. 12s1–3d, pp. 287–295.

f273g Dekker, T., Takken, W., Knols, B. G. J., Bouman, E., van de Laak, SBever, A., and Huisman, P. W. T., 1998, “Selection of Biting Sites oHuman Host by Anopheles Gambiae, An-Arabiensis andQuadriannulatus,” Entomol. Exper. App., 87s3d, pp. 295–300.

f274g Knols, B. G. J., Takken, W., Cork, A., and DeJong, R., 1997, “OdMediated, Host-Seeking Behavior of Anopheles Mosquitoes: A Newproach,” Ann. Trop. Med. Parasitol.,91, pp. S117–S118.

f275g Takken, W., 1996, “Synthesis and Future Challenges: The Response oquitoes to Host Odours,” Ciba Found Symp.,200, pp. 302–312.

f276g McIver, S. B., 1982, “Sensilla of MosquitoessDiptera:Culicidaed,” J. Med.Ethics, 19, pp. 489–535.

f277g Sutcliffe, J. F., 1994, “Sensory Bases of Attractancy: Morphology of Mquito Olfactory Sensilla—A Review,” J. Am. Mosq. Control Assoc.,10s2d,Part 2, pp. 309–315.

f278g Bidlingmayer, W. L., Day, J. F., and Evans, D. G., 1995, “Effect of W
Velocity on Suction Trap Catches of Some Florida Mosquitos,” J. Am. Mosq
216 / Vol. 127, MARCH 2005

,

-to

-

l.

,

ei.

ts

fi-

nd

d

s

er,

by

.,ood

e

-

--

s-

-

Control Assoc.,11s3d, pp. 295–301.f279g Békésy, G. v., 1964, “Olfactory Analogue to Directional Hearing,” J. A

Physiol., 19s3d, pp. 369–373.f280g Kobal, G., Van Toller, S., and Hummel, T., 1989, “Is there Directional Sm

ing?,” Experientia,45s2d, pp. 130–132.f281g Stoddart, D. M., 1979, “External Nares and Olfactory Perception,” Expe

tia, 35, pp. 1456–1457.f282g Kajiura, S. M., Forni, J. B., and Summers, A. P., 2005, “Olfactory Morp

ogy of Carcharhinid and Sphyrnid Sharks-Does the Cephalofoil ConSensory Advantage?,” J. Morphol., in press.

f283g Vanneste, E., and Geise, H. J., 2003, “Commercial Electronic Nose Iments,” inHandbook of Machine Olfaction, T. C. Pearce et al., eds., WileNew York, pp. 161–179.

f284g Snopok, B. A., and Kruglenko, I. V., 2002, “Multisensor Systems for Chcal Analysis: State-of-the-Art in Electronic Nose Technology andTrends in Machine Olfaction,” Thin Solid Films,418s1d, pp. 21–41.

f285g Yinon, J., 2003, “Detection of Explosives by Electronic Noses,” AChem., 75s5d, pp. 99A–105A.

f286g Li, J., 2000, “The Cyranose Chemical Vapor Analyzer,” Sensors,17s8d, pp.56–60.

f287g Cumming, C., Aker, C., Fisher, M., Fox, M., IaGrone, M., Reust, D., Rley, M., Swager, T., Towers, E., and Williams, V., 2001, “Using Novel Frescent Polymers As Sensory Materials for Above-Ground Sensing of Ccal Signature Compounds Emanating From Buried Landmines,” IEEE TGeosci. Remote Sens.,39s6d, pp. 1119–1128.

f288g Albert, K. J., Myrick, M. L., Brown, S. B., James, D. L., Milanovich, F.and Walt, D. R., 2003, “Field-Deployable Sniffer for 2,4-Dinitrotoluenetection,” Environ. Sci. Technol.,35s15d, pp. 3193–3200.

f289g Staples, E. J., 2001, “The ZNose,” Sensors,18s6d, pp. 92–97.f290g Gelperin, A., and Hopfield, J. J., 2002, “Electronic and Computationa

faction,” ACS Symposium Series,825, pp. 289–317.f291g Gelperin, A., Oct. 7, 1997, “Olfactory Sensor Identification System

Method,” U.S. Patent 5,675,070.f292g White, J. E., Waggoner, L. P., and Kauer, J. S., 2004, “Explosives and

mine Detection Using an Artificial Olfactory System,”Proceedings SPISecurity Defense Conference, Orlando, Proc. SPIE,5415, Paper 35, pp. 521532.

f293g Kauer, J. S., and White, J. E., 2003, “Representation of Odor Informatthe Olfactory System: From Biology to an Artificial Nose,” inSensors anSensing in Biology and Engineering, Springer-Verlag, Vienna, pp. 305–32

f294g White, J. E., Mall, S., and Kauer, J. S., 2002, “Using Biology to GDevelopment of an Artificial Olfactory System,” inNeurotechnology for Biomimetic Robots, J. Ayers, J. L. Davis, and A. Rudolph, eds., MIT, CambridMA.

f295g Eiceman, G. A., 2002, “Ion-Mobility Spectrometry as a Fast MonitoChemical Composition,” Trends Analyt. Chem.,21s4d, pp. 259–275.

f296g Hill, H. H., Siems, W. F., StLouis, R. H., and McMinn, D. G., 1990, “Mobility Spectrometry,” Anal. Chem.,62s23d, pp. A1201–A1209.

f297g Linker, K. L., Adkins, D. R., Mowry, C. D., Trudell, D. E., Pfeifer, K. BSanchez, R. C., and Rohde, S. B., 2001, “Micro-Explosives Detection SmHound™,” in Proceedings of the 3rd FAA International SymposiumExplosives Detection and Aviation Security Technologies, Atlantic City, NJ,U.S. Federal Aviation Administration.

f298g Call, P., Kenning, V., and Moler, C., 2001, “BioCapture Air Sampler,” htwww.mesosystems.com/

f299g McGown, J. B., Bromberg, E. A., and Noble, L. W., Mar. 20, 1990, “VaSampling Probe,” U.S. Patent 4,909,090.

f300g Carroll, A. L., Miskolczy, G., Fraim, F. W., Achter, E. K., and Lieb, D.June 23, 1992, “Hand-Held Sample Gun for Vapor Collection,” U.S. P5,123,274.

f301g Kolb, B., and Ettre, L. S., 1997,Static Headspace-Gas ChromatograpTheory and Practice, Wiley-VCH, New York.

f302g Davidson, W. R., and Stott, W. R., 1994, “Synergy or Dichotomy—VaporParticle Sampling in the Detection of Contraband,”Proceedings SPIE Coference on Substance Detection Systems, Innsbruck, Austria, Proc. SPI2092, pp. 108–119.

f303g Fainberg, A., 1992, “Explosives Detection for Aviation Security,” Scie255s5051d, pp. 1531–1537.

f304g Steinfeld, J. I., and Wormhoudt, J., 1998, “Explosives Detection: A Chalfor Physical Chemistry,” Annu. Rev. Phys. Chem.,49, pp. 203–232.

f305g Bender, E., Hogan, A., Leggett, D., Miskolczy, G., and Macdonald, S., 1“Surface Contamination by TNT,” J. Forensic Sci.,37s6d, pp. 1673–1678.

f306g Brown, R. H., and Monteith, L. E., 2001, “Gas and Vapor Sample Cotors,” in Air Sampling Instruments for Evaluation of Atmospheric Contnants, 9th Ed., American Conference of Governmental Industrial HygienCincinnati, OH, pp. 415–455.

f307g Anonymous, 1989, “Part I—General Techniques,” inMethods of Air Sampling and Analysis, 3rd Ed., J. P. Lodge, ed., Lewis Publishers, New York3–58.

f308g Brown, H. R., and Purnell, C. J., 1979, “Collection and Analysis of TOrganic Vapor Pollutants in Ambient Atmospheres,” J. Chromatogr.,178, pp.97–80.

f309g Bender, F., Barie, N., Romoudis, G., Voigt, A., and Rapp, A., 2003, “Deopment of a Preconcentration Unit for a SAW Sensor Micro Array anUse for Indoor Air Quality Monitoring,” Sens. Actuators B,93s1–3d, pp.

. 135–141.


ffi-f, Ntion

, F.S.

ection,

g Ai

lowHyg

cup

Pra.

Loc

athion

oro

-

s,” iPar

andmbe

ng f

Reyrma

,” in,

ted

-Lintros

r Extion,

Effilant

Fol

J.,ffingput

s oan

nina--tic

t Oan

al-

ce of

“Thef

tra-

kes,motedcurity

s of

57.s for

odsState

nolo-eport

tion,”Jose,

ey of

nter-

sal of

Re-

ting,

, “Theody,”blemJ,

ann,ives

ium-

90,tent

tical

on thetent

eopleSub-

ethodnger-

l forpo-r

sferuild.

Air-uman

entalEng.ivesecu-

,

nnelon

c,

ent.

hier,reen-

f310g Yinon, J., 1999,Forensic and Environmental Detection of Explosives, Wiley,New York.

f311g Fraim, F. W., Achter, E. K., Carroll, A. L., and Hainsworth, E., 1992, “Ecient Collection of Explosive Vapors, Particles and Aerosols,”Proceedings othe 1st International Symposium on Explosives Detection Technology1991, S. M. Khan, ed., Report DOT/FAA/CT-92/11 U. S. Federal AviaAdministration, Atlantic City, NJ, pp. 559–565.

f312g Achter, E. K., Carroll, A. L., Roundbehler, D. P., Fine, D. H., and FraimW., Mar. 3, 1992, “Vapor Collector/Desorber With Metallic Ribbon,” UPatent 5,092,157.

f313g Linker, K. L., 2001, “Large-Volume Sampling and Preconcentration,” inPro-ceedings of the 3rd FAA International Symposium on Explosives Detand Aviation Security Techniques, U. S. Federal Aviation AdministratioAtlantic City, NJ.

f314g Dravnieks, A., 1966, “Odors As Signatures,” New Sci.,31, pp. 622–624.f315g Aaberg, C. P. N., Aug. 23, 1977, “Method and Apparatus for Exhaustin

from a Limited Zone,” U.S. Patent 4,043,257.f316g Kulmala, I., 2000, “Experimental Validation of Potential and Turbulent F

Models for a Two-Dimensional Jet-Enhanced Exhaust Hood,” Am. Ind.Assoc. J.,61sMar./Apr.d, pp. 183–191.

f317g Burton, D. J., 2000, “Practical Applications of an Aaberg Hood,” OcHealth Saf.,69s11d, pp. 37–39.

f318g Heinsohn, R. J., and Cimbala, J. M., 2002, “Present Local Ventilationtice,” in Indoor Air Quality Engineering, Marcel Dekker, New York, pp437–515.

f319g Rueeg, T., Gubler, D., Sprecher, P., Moser, A., and Weber, R., 2004, “Capture of Contaminants,” ASHRAE J.,46s1d, pp. 26–29.

f320g Hunt, G. R., and Ingham, D. B., 1996, “Long Range Exhaustion—A Mematical Model for the Axisymmetric Airflow of a Local Exhaust VentilatHood Assisted by a Turbulent Radial Jet,” Ann. Occup. Hyg.,40s2d, pp.171–196.

f321g Cant, R., Castro, I., and Walklate, P., 2002, “Plane Jets Impinging on PWalls,” Exp. Fluids, 32s1d, pp. 16–26.

f322g Anonymous, 2001, “Air curtains,” inIndustrial Ventilation Design Guidebook, Academic, San Diego, pp. 553–571.

f323g Stenger, J. B., and Bajura, R. A., 1984, “Deposition in Sampling TubeAerosols: Science, Technology and Industrial Applications of Airborneticles, Proceedings of the First International Aerosol Conference, Elsevier,New York, pp. 175–178.

f324g Falcitelli, M., Benassi, A., Di Francesco, F., Domenici, C., Marano, L.,Pioggia, G., 2002, “Fluid Dynamic Simulation of a Measurement Chafor Electronic Noses,” Sens. Actuators B,85s1–2d, pp. 166–174.

f325g Laibach, E., 1937, “Das Geruchsorgan des AalssAnguilla Vulgarisd in SeinenVerschiedenen Entwicklungsstadien,” Zoologische Jahrbucher, AbterluAnatomie und Ontogenie der Tiere,63, pp. 37–72.

f326g Rae, W. H., and Pope, A. Y., 1984,Low-Speed Wind Tunnel Testing, Wiley,New York.

f327g Mälhammar, A., 2004, “A Method for Comparing Heat Sinks Based onnolds Analogy,”Proceedings of the 10th International Workshop on TheInvestigation of ICs and Systems, Sophia Antipolis, France, pp. 229–234.

f328g Eckert, E. R. G., and Drake, R. M., 1959,Heat and Mass Transfer, McGraw-Hill, NY.

f329g Garimella, S. V., 1996, “Enhanced Air Cooling of Electronic equipmentAir Cooling Technology for Electronic Equipment, S. J. Kim and S. W. Leeeds., CRC, Boca Raton, FL, pp. 173–202.

f330g Settles, G. S., 2006, “Fluid Mechanics and Homeland Security,” submitAnnu. Rev. Fluid Mech., 38.

f331g Carranza, J. E., Fisher, B. T., Yoder, G. D., and Hahn, D. W., 2001, “OnAnalysis of Ambient Air Aerosols Using Laser-Induced Breakdown Speccopy,” Spectrochim. Acta, Part B,56s6d, pp. 851–864.

f332g Myers, L. J., 1992, “The Dog-Handler Team as a Detection System foplosives: A Tail to be Told,”Proceedings of the Conference on the AviaSecurity Problem and Related Technologies, W. H. Makky, ed., SPIE CR42San Diego, pp. 93–103.

f333g Furton, K. G., and Myers, L. J., 2001, “The Scientific Foundation andcacy of the Use of Canines As Chemical Detectors for Explosives,” Ta54s3d, pp. 487–500.

f334g Gazit, I., and Terkel, J., 2003, “Explosives Detection by Sniffer Dogslowing Strenuous Physical Activity,” App. Animal Behav. Sci., 81s2d, pp.149–161.

f335g Gazit, I., Lavner, Y., Bloch, G., Azulai, O., Goldblatt, A., and Terkel,2003, “A Simple System for the Remote Detection and Analysis of Sniin Explosives Detection Dogs,” Behav. Res. Methods Instrum. Com35s1d, pp. 82–89.

f336g Rooney, N. J., Bradshaw, J. W. S., and Almey, H. E., 2004, “AttributeSpecialist Search Dogs-A Questionnaire Survey of UK Dog HandlersTrainers,” J. Forensic Sci.,49s2d, pp. 300–306.

f337g Ternes, J. W., and Prestrude, A. M., 1992, “Integration of the Human, CaMachine Interface for Explosives Detection,”Proceedings of the 1st Interntional Symposium on Explosive Detection Technology, S. M. Khan, ed., Report DOT/FAA/CT-92/11 U. S. Federal Aviation Administration, AtlanCity, NJ, pp. 891–902.

f338g Settles, G. S., 1999, “Schlieren and Shadowgraph Imaging in the Greadoors,”Proceedings of the 2nd Pacific Symposium on Flow VisualizationImage Processing, S. Mochizuki, ed., Honolulu, Pacific Center of ThermFluids Engineering, Tokyo.

f339g Fine, D. H., Fraim, F. W., MacDonald, S. J., and Thrash, K. M., Dec. 27


ov.

.

n

r

.

.

c-

al

-

us

n-

r

ur

-l

to

e-

-

-a,

-

.,

fd

e,

ut-d

1994, “Method and System for Sampling and Determining the PresenCompounds in Containers,” U.S. Patent 5,376,550.

f340g Davidson, W. R., Stott, W. R., Akery, A. K., and Sleeman, R., 1992,Role of Mass Spectrometry in the Detection of Explosives,”Proceedings othe 1st International Symposium on Explosive Detection Technology, S. M.Khan, ed., Report DOT/FAA/CT-92/11, U. S. Federal Aviation Administion, Atlantic City, NJ, pp. 663–667.

f341g Cartwright, P. A., Alcock, N. J., Groves, D. M., Almey, H. E., and BrooM. D., 2001, “Development of Performance Evaluation Tubes for ReAir Sampling for Canine OlfactionsRASCOd,” in Proceedings of the 3rFAA International Symposium on Explosives Detection and Aviation SeTechnologies, U. S. Federal Aviation Administration, Atlantic City, NJ.

f342g Lord, D., 2004, “RASCO: A New Cargo Screening Methodology ComeAge,” Aviation Security International,9s4d, p. 34.

f343g Jenkins, A., Mar. 9, 1976, “Inspection Apparatus,” U.S. Patent 3,942,3f344g Ornath, F., and Beuchler, S. S., Aug. 24, 1999, “Method and Apparatu

Sampling Contaminants,” U.S. Patent 5,942,699.f345g Smith, D. J., 2005, “The Internal Airflow in Cargo Containers and Meth

to Sample it for Chemical Trace Detection,” M.S. Thesis, PennsylvaniaUniversity, State College PA.

f346g Anonymous, 1996, “Airline Passenger Security Screening—New Techgies and Implementation Issues,” US National Research Council RNMAB-482-1, U. S. National Academy of Science, Washington, DC.

f347g Anonymous, 2002, “People Screening Weapon & Explosives DetecMarket Report 2003–2010, Homeland Security Research Corp., SanCA.

f348g Hallowell, S. F., 2001, “Screening People for Illicit Substances: a SurvCurrent Portal Technology,” Talanta,54s3d, pp. 447–458.

f349g Dravnieks, A., 1975, “Evaluation of Human Body Odor: Methods and Ipretations,” J. Soc. Cosmet. Chem.,26s11d, pp. 551–571.

f350g Bethune, D. W., Blowers, R., Parker, M., and Pask, E. A., 1965, “DisperStaphylococcus Aureusby Patients and Surgical Staff,” Lancet,318, pp.480–483.

f351g Whyte, W., Vesley, D., and Hodgson, R., 1976, “Particle Dispersion inlation to Operating Room Clothing,” J. Hyg.sLondd, 76, pp. 367–378.

f352g Whyte, W., 2001,Cleanroom Technology: Fundamentals of Design, Tesand Operation, Wiley, New York.

f353g Settles, G. S., Gowadia, H. A., Strine, S. B., and Johnson, T. E., 1996Natural Aerodynamic Sampling of Trace Explosives From the Human BProceedings of the 2nd FAA Conference on the Aviation Security Proand Related Technologies, W. H. Makky, ed., SPIE CR42, Atlantic City, Npp. 65–70.

f354g Settles, G. S., Ferree, H. C., Tronosky, M. D., Moyer, Z. M., and McGW. J., 2001, “Natural Aerodynamic Portal Sampling of Trace ExplosFrom the Human Body,” inProceedings 3rd FAA International Symposon Explosive Detection and Aviation Security, U. S. Federal Aviation Administration, Atlantic City, NJ.

f355g Achter, E. K., Burke, E. J., Miskolczy, G., and Sonin, A. A., Mar. 20, 19“Walk-in Inspection Apparatus for Producing Air Samples,” U.S. Pa4,909,089.

f356g Linker, K. L., Hannum, D. W., and Conrad, F. J., June 22, 1999, “VerFlow Chemical Detection Portal,” U.S. Patent 5,915,268.

f357g Settles, G. S., June 13, 2000, “Chemical Trace Detection Portal BasedNatural Airflow and Heat Transfer of the Human Body,” U.S. Pa6,073,499.

f358g Davies, J. H., Apr. 23, 2002, “Apparatus and Method for Screening Pand Articles to Detect and/or to Decontaminate with Respect to Certainstances,” U.S. Patent 6,375,697.

f359g Phares, D. J., Holt, J. K., Smedley, G. T., and Flagan, R. C., 2000, “Mfor Characterization of Adhesion Properties of Trace Explosives in Fiprints and Fingerprint Simulations,” J. Forensic Sci.,45s4d, pp. 774–784.

f360g Neudorfl, P., McCooeye, M. A., and Elias, L., 1993, “Testing ProtocoSurface-Sampling Detectors,”Proceedings of the 4th International Symsium on the Analysis and Detection of Explosives, J. Yinon, ed., KluweAcademic, New York, pp. 373–384.

f361g Sorensen, D. N., and Voight, L. K., 2003, “Modelling Flow and Heat TranAround a Seated Human Body by Computational Fluid Dynamics,” BEnviron., 38s6d, pp. 753–762.

f362g Murakami, S., Kato, S., and Zeng, J., 2000, “Combined Simulation offlow, Radiation and Moisture Transport for Heat Release From a HBody,” Build. Environ., 35s6d, pp. 489–500.

f363g Craven, B. A., and Settles, G. S., 2005, “A Computational and ExperimInvestigation of the Human Thermal Plume,” submitted to J. Biomech.

f364g Schellenbaum, R. L., 1987, “Air Flow Studies for Personnel ExplosScreening Portals,”Proceedings of the 1987 Carnahan Conference on Srity Technology: Electronic Crime Countermeasures, Lexington, KY, IEEENew York, pp. 101–105.

f365g Hobbs, J. R., and Conde, E. P., 1993, “Analysis of Airflows in PersoScreening Booths,”Proceedings of the Fourth International Symposiumthe Analysis and Detection of Explosives, J. Yinon, ed., Kluwer AcademiNew York, pp. 437–453.

f366g Daws, L. F., 1970, “Movement of Airstreams Indoors,” J. Inst. Heat. VEng., 37s2d, pp. 241–253.

f367g Parmeter, J. E., Linker, K. L., Rhykerd, C. L., Hannum, D. W., and BoucF. A., 1998, “Explosives Detection Portal for High-Volume Personnel Scing,” in Enforcement and Security Technologies, Proc. SPIE,3575, pp. 384–

, 391.

MARCH 2005, Vol. 127 / 217

rtal,

room,

ree-

on oor

gi-

Stued t

ConAm

a

theure,

on

ind

of arsit

Tracson

.

f Laque

.

.e

Fa

an

ey,re o

. Ailled

99ion

fors,” iMin

tro-ectition

iat.

enc

. J.,

e and

. J.,

hizo-

har-atol.,

ofsoseshing-

andn iniumds.,

Ionical

sionscup.

. M.,itivity

thrC, pp.

. J.,aserma-

ical

sticst

-

and

Leak

Pipe-

Gas

e,

hen,mics,

f368g Linker, K. L., and Brusseau, C. A., Jan. 1, 2002, “Target Detection PoU.S. Patent 6,334,365.

f369g Whyte, W., and McGeorge, D., 1994, “Evaluating the Effects of a CleanAir Shower on Particle Dispersion,” Microcontamination,12s9d, pp. 39–4169.

f370g Batill, S. M., and Hoffman, J. J., 1986, “Aerodynamic Design of ThDimensional Subsonic Wind Tunnel Inlets,” AIAA J.,24s2d, pp. 268–269.

f371g Settles, G. S., and McGann, W. J., 2001, “Potential for Portal DetectiHuman Chemical and Biological Contamination,” inAerosense 2001: SensTechnology and Application, SPIE-International Society for Optical Enneering4378, Paper 01.

f372g Edge, B. A., Paterson, E. G., and Settles, G. S., 2005, “Computationalof the Wake and Contaminant Transport of a Walking Human,” submittASME J. Fluids Eng.

f373g Rodes, C. E., Kamens, R. M., and Wiener, R. W., 1995, “Experimentalsiderations for the Study of Contaminant Dispersion Near the Body,”Ind. Hyg. Assoc. J.,56s6d, pp. 535–545.

f374g Kim, T., and Flynn, M. R., 1991, “Airflow Pattern Around a Worker inUniform Freestream,” Am. Ind. Hyg. Assoc. J.,52s7d, pp. 287–296.

f375g Flynn, M. R., and Miller, C. T., 1991, “Discrete Vortex Methods forSimulation of Boundary-Layer Separation Effects on Worker ExposAnn. Occup. Hyg.,35s1d, pp. 35–50.

f376g Flynn, M. R., and Ljungqvist, B., 1995, “A Review of Wake EffectsWorker Exposure,” Ann. Occup. Hyg.,39s2d, pp. 211–221.

f377g Murakami, S., Zeng, J., and Hayashi, T., 1999, “CFD Analysis of WEnvironment Around a Human Body,” J. Wind. Eng. Ind. Aerodyn.,83, pp.393–408.

f378g Moyer, Z. M., 2003, “The Human Aerodynamic Wake and the DesignPortal to Sample It,” M.S. thesis, Pennsylvania State University, UnivePark, PA.

f379g Craven, B. A., and Settles, G. S., 2005, “The Design of a ChemicalDetection Portal That Samples the Aerodynamic Wake of a Walking Perin preparation.

f380g Anonymous, 1998,Hidden Killers: The Global Landmine Crisis, U.S. Deptof State Publication 10575.

f381g Andersson, N., Dasousa, C. P., and Paredes, S., 1995, “Social Cost oMines in 4 Countries—Afghanistan, Bosnia, Cambodia, and MozambiBr. Med. J., 311s7007d, pp. 718–721.

f382g Ashley, S., 1996, “Searching for Land Mines,” Mech. Eng.sAm. Soc. MechEng.d, 118s4d, pp. 62–67.

f383g Strada, G., 1996, “The Horror of Land Mines,” Sci. Am.,278s5d, pp. 40–45f384g Pamula, V. K., 2003, “Detection of Explosives,” inHandbook of Machin

Olfaction, T. C. Pearce et al., eds., Wiley, New York, pp. 547–580.f385g Phelan, J., and Webb, S. W., 1998, “Simulation of the Environmental

and Transport of Chemical Signatures From Buried Landmines,”Proceed-ings of the 1999 Detection and Remediation Technologies for MinesMinelike Targets III, Proc. SPIE,3392s1d, pp. 509–520.

f386g George, V., Jenkins, T. F., Leggett, D. C., Cragin, J. H., Phelan, J., Oxland Pennington, J., 1999, “Progress on Determining the Vapor SignatuBuried Land Mine,” Proceedings of the SPIE Aerosense Conference, Or-lando, FL, Proc. SPIE,3710, pp. 258–269.

f387g Jenkins, T. F., Leggett, D. C., Miyares, P. H., Walsh, M. E., Ranney, TCragin, J. H., and George, V., 2001, “Chemical Signatures of TNT-FLand Mines,” Talanta,54s3d, pp. 501–513.

f388g Morgan, J. S., Bryden, W. A., Miragliotta, J. A., and Aamodt, L. C., 19“Improved Detection of Explosive Residues by Laser Thermal DesorptJohns Hopkins APL Tech. Dig.,20s3d, pp. 389–395.

f389g Fair, R. B., Pollack, M., and Pamula, V. K., 1998, “MEMS DevicesDetecting the Presence of Explosive Material Residues in Mine FieldProceedings of the 1999 Detection and Remediation Technologies forand Minelike Targets III, Proc. SPIE,3392s1d, pp. 409–417.

f390g Fisher, M., Cumming, C., IaGrone, M., and Taylor, R., 1998, “An Elecstatic Particle Sampler and Chemical Sensor System for Landmine Detby Chemical Signature,”Proceedings of the 1998 Detection and RemediaTechnologies for Mines and Minelike Targets III, Proc. SPIE,3392s1d, pp.565–574.

f391g Morrocco, J. D., 2000, “Field Trials in Kosovo Test Mine Detection,” AvWeek Space Technol.,156, p. 38.

f392g Turner, A. P. F., 2000, “Biosensors—Sense and Sensitivity,” Sci

218 / Vol. 127, MARCH 2005

”

f

dyo

-.

”

y

e,”

nd,”

te

d

J.,f a

.,

,,”

nes

on

e,

290s5495d, pp. 1315–1317.f393g Cole, L. A., 2003,The Anthrax Letters: A Medical Detective Story, Joseph

Henry Press, Washington DC.f394g Pope, C., 1928, “The Diagnostic Art of Smelling,” Am. J. Med.,34, pp.

651–653.f395g Hayden, G. F., 1980, “Olfactory Diagnosis in Medicine,” Postgrad Med

67s4d, pp. 110–118.f396g Tassopoulos, C. N., Barnett, D., and Fraser, T. R., 1969, “Breath-Aceton

Blood-Sugar Measurements in Diabetes,” Lancet,1s1282d, pp. 1286.f397g McSherry, J., 1986, “Sniffing Out the Diagnosis,” Can. Med. Assoc

135s10d, pp. 1070.f398g Smith, K., Thompson, G. F., and Koster, H. D., 1969, “Sweat in Sc

phrenic Patients: Identification of the Odorous Substance,” Science,166, pp.398–399.

f399g Distelheim, I. H., and Dravnieks, A., 1973, “A Method for Separating Cacteristics of Odors in Detection of Disease Processes 1,” Int. J. Derm12s4d, pp. 241–244.

f400g Di Natale, C., and D’Amico, A., 2001, “The Electronic Nose AnalysisBreath As Potential Diagnostic Tool: the Case of Lung Cancer,”Proceedingof the Eighth International Symposium on Olfaction and the Electronic N,J. R. Stetter and W. R. Penrose, eds., The Electrochemical Society, Waton DC, pp. 48–53.

f401g Krüger, U., Körber, R., Ziegler, K.-H., Koronczi, I., Nachnani, S.,Goschnick, J., 2001, “Electronic Noses Towards Practical ApplicatioMedical Diagnostics,”Proceedings of the Eighth International Symposon Olfaction and the Electronic Nose, J. R. Stetter and W. R. Penrose, eThe Electrochemical Society, Washington DC, pp. 62–64.

f402g Lawrence, A. H., Nanji, A. A., and Taverner, J., 1988, “Skin-SniffingMobility Spectrometric Analysis—A Potential Screening Method in ClinToxicology,” J. Clin. Lab Anal,2s2d, pp. 101–107.

f403g Batterman, S. A., Franzblau, A., and Zhou, N.-Z., 1996, “Airborne Emisat Skin Surfaces: a Potential Biological Exposure Index,” Int. Arch. OcEnviron. Health,68, pp. 268–274.

f404g Stetter, J. R., Penrose, W. R., Kubba, S., Kocka, F., McEntegart, CRoberts, R. R., and Iademarco, M. F., 2001, “Enhancement of the Sensof the Electronic Nose for Bacterial Applications,”Proceedings of the EighInternational Symposium on Olfaction and the Electronic Nose, J. R. Stetteand W. R. Penrose, eds., The Electrochemical Society, Washington D54–61.

f405g Roller, C., Namjou, K., Jeffers, J. D., Camp, M., Mock, M., McCann, Pand Grego, J., 2002, “Nitric Oxide Breath Testing by Tunable-Diode LAbsorption Spectroscopy: Application in Monitoring Respiratory Inflamtion,” Appl. Opt., 41s28d, pp. 6018–6029.

f406g Pavlou, A. K., and Turner, A. P. F., 2000, “Sniffing Out the Truth: ClinDiagnosis Using the Electronic Nose,” Clin. Chem. Lab. Med.,38s2d, pp.99–112.

f407g Persaud, K. C., Pisanelli, A. M., and Evans, P., 2003, “Medical Diagnoand Health Monitoring,” inHandbook of Machine Olfaction, T. C. Pearce eal., eds., Wiley, New York, pp. 445–460.

f408g Settles, G. S., 2004, “The Penn State Full-Scale Schlieren System,”Proceedings of the 11th International Symposium on Flow Visualization, T. J. Muellerand I. Grant, eds., Notre Dame University, IN.

f409g Kaminski, R. K., 1980, “Leak Detection: A Route to ConservationSafety,” Proc., Am. Pet. Inst., Sect. 3,59, pp. 37–41.

f410g Peale, R. E., and Summers, P. L., 1996, “Zebra Schlieren Optics forDetection,” Appl. Opt.,35s22d, pp. 4518–4521.

f411g Settles, G. S., 1999, “Imaging Gas Leaks by Using Schlieren Optics,”line and Gas Journal,226s9d, pp. 28–30.

f412g Mcrae, T. G., and Kulp, T. J., 1993, “Backscatter AbsorptionImaging—A New Technique for Gas Visualization,” Appl. Opt.,32s21d, pp.4037–4050.

f413g Kher, U., 2000, “Electronic Noses Sniff Out a Market or Two,” Tim155s11d, pp. 64–65.

f414g Aston, C., 2001, “Biological Warfare Canaries,” IEEE Spectrum,38s10d, pp.35ff.

f415g Olender, T., Fuchs, T., Linhart, C., Shamir, R., Adams, M., Kalush, F., KM., and Lancet, D., 2004, “The Canine Olfactory Subgenome,” Geno83s3d, pp. 361–372.


sniffers: fluid-dynamic sampling for olfactory trace ... · sniffers is the principal aim of this...

Documents