..

UCRbJc-127526PREPRINT

Unexploded Ordnance Detection Using ImagingGiant Magnetoresistive (GMR) Sensor Arrays

Alison Chaiken

This paper was prepared for submittal to theUXO Forum 97Nashville, TN

May 27-30,1997

May 6,1997

DISCLAIMER

This document was prepared as an account of work sponsored by an agency ofthe United States Government. Neither the United States Government nor theUniversity of California nor any of their employees, makes any warranty, expressor implied, or assumes any legal liability or responsibility for the accuracy,completeness, or usefulness of any information, apparatus, product, or processdisclosed, or represents that its use would not infringe privately owned rights.Reference herein to any specific commercial product, process, or service by tradename, trademark, manufacturer, or otherwise, does not necessarily constitute orimply its endorsement, recommendation, or favoring by the United StatesGovernment or the University of California. The views and opinions of authorsexpressed herein do not necessarily state or reflect those of the United StatesGovernment or the University of California, and shall not be used for advertisingor product endorsement purposes.

UNEXPLODEDORDNANCEDETECTIONUSINGIMAGINGGIANTMAGNETORESISTIVE(GMR)SENSORARWYS

Alison ChaikenMailstop L-350

Lawrence Livermore NationalLabLivermore, CA 94550

chaiken@llnl.govhttp#/www.wsrcc.com/alison/magnetism.html

(510) 422-7129

ABSTRACT

False positive detections account for a greatpart of the expense associated withunexploded ordnance (UXO) remediation.Presently fielded systems like pulsedelectromagnetic induction systems andcesium-vapor magnetometers are able todistinguish between UXO and other metallicground clutter only with tlfficul~. Thediscovery of giant magnetoresistance(GMR) has led to the development of a newgeneration of integrated-circuit magneticsensors that are far more sensitive thanpreviously available room-temperature-operation electronic devices. The sma.11sizeof GMR sensors makes possible theconstruction of array detectors that can beused to image the flux emanating from aferrous object or fkom a non-ferrous objectwith eddy currents imposed by an externalcoil. The purpose of a GMR-based imagingdetector would be to allow the operator toeasily distinguish between UXO and benignobjects (like shrapnel or spent bullets) thatlitter formerly used defense sites (FUDS).

In order to demonstrate the potential of aGMR-based imaging technology, a crudemagnetic imaging system has beenconstructed using commercially availablesensors. The ability to roughly determinethe outline and disposition of magneticobjects has been demonstrated.Improvements to the system which arenecessary to make it into a high-performance UXO detector are outlined.

INTRODUCTION

Many techniques are in use or have beenproposed for use as UXO detectors. The

two most commonly employed technologiesare electromagnetic induction detection andfluxgate magnetometry. While time-domain analysis of inductive signals hasbeen suggested as a way to differentiatebetween hazardous aud benign types ofburied materkd, neither the inductiondetector nor the fluxgate magnetometer maybe engineered to produce an image ofpotential UXO objects. The success ofimaging technologies based on arrays ofdetectors like forward-looking infraredcameras for infrared ~get identification andcharge-coupled device video cameras forconsumer applications suggests that thesensor-array paradigm is worth exploring forUXO detection as well. Neither theelectromagnetic induction nor fluxgatemagnetometry methods is well-suited forincorporation into a detector array sincethese sensors are bulky in size. Newlydeveloped GMR sensors, on the other hand,are now available now in integrated circuitform. These sensors are attractive for avariety of applications because of their highsensitivity (over ten times greater thag Hallsensors), room-temperature operation(unlike SQUID magnetometers) andmoderate cost (currently $5 each in smallquantities).

Physics of the Giant Magnetoresistance

The giant magnetoresistance (GMR) effectwas discovered in France in 1988, but it hasbeen widely investigated by USinvestigators. ~aibich, 1988] As illustratedin FigCure1, GMR is a very large change inelectrical resistance that is observed in aferromagnet/pararnagnet multilayer structurewhen the relative orientations of themagnetic moments in alternateferromagnetic layers change as a function of .

1

applied field. The basis of the GMR isthe dependence of the electrical resistivitv ofelectrons in a magnetic metal on the “direction of the electron spin, either parallelor arrtiparallel to the magnetic moment ofthe films (indicated by heavy arrowsbelow). Electrons which have a parallelspin undergo less scattering and thereforehave a lower resistance. When themoments of the magnetic layers (NiFe inFig. 1) are antiparallel at low field, there areno electrons which have a low scatteringrate in both magnetic layers, causing anincreased resistance. At applied magneticfields where the moments of the magneticlayers are aligned, electrons with their spinsparallel to these moments pass freelythrough the solid, lowering the electricalresistance. The resistance of the structureis therefore proportional to the cosine of theangle between the magnetic moments in

/ Cu I

I substrate I

paramagnetic metal causes a-zero-fieldantipsrallel alignment which can beovercome by a high applied tield.~inasch,1989] The magnitude of the GMR effect cambe surprisingly large, up to 80% at roomtemperature in Co/Cu multilayers asreported by workers at Srmyo.[KaIIo, 1993]However, tire fields needed to saturateCo/Cu multilayers are too large for sensoraPPli~tions. Otier Smit.ilayersare designedto have an antipsrallel state in a limitedaPplicd field rmge by alternatingferromagnetic layers (Co and Fe layersinstead of two NiFe layers) with differentintrinsic switching fields. [Chaiken, 199 I]Outputs of GMRs can be as large as 12% at20 Oe in film form, with slightly lowersensitivity found in rnicrofabricateridevices. [Arsthony, 1994] The NVE sensorsused in this study have an output of only0.3% at 15 Oe, so considerable improvement

Cu

I substrate IZero-Field Hkzh Resistance State Hi~h-Field Low Resistance State

F@re 1. The Gkurt Magnetoresistarrce effect is due to the large difference in electricalresistance between two magnetic states of a metallic mukilayer film.

adjacent magnetic layers. [Chaiken, 1990]

The occurrence of the GMR effect dependson the ability of the applied magnetic toswitch the relative orientation of themagnetic moments back and forth betweenthe parallel and arrtiparallel states. In somemultilayers a quantum-mechanical interlayerexchange coupling across Cu or another

is expected in the future.

GMR sensors have recently been evaluatedfor use in geophysical exploration and foundto have a noise floor of 0.1 to 1.0 nT in anunshielded, unfiltered system. [McGlone,1997] This sensitivity is comparable to anelectromagnetic induction system, althoughnot at the level of cesium-vapormagnetometer systems. A practic~ UXO .

2

Full array

f

\

Corners

Others

Figure 2. An illustration of the layout of the 5x5 GMR sensor array. Each white square in the“full may” drawing on the left represents one NVE NVS5B 15 sensor. The arrows on thisdrawing indicate the orientation of the axis of sensitivity for each sensor. The right side of thefigure shows how the outputs of the elements are split up in the images that follow.

system in the end is expected to incorporatea variety of sensor types integrated into asingle package so that maximum sensitivityand imaging capability wiff be available tooperate in concert.

The impact of GMR array UXO detectors onDOD site remediation activities ispotentially great. The inspection of falsepositives during cleanup of contaminatedareas adds greatly to the cost and duration ofsite remediation. Typicafly, 50 to 60pounds of scrap metal are recovered for eachordnance item found using presenttechnology. An easy-to-use imaging UXOdetection system would allow a relativelyinexperienced user to rapidly distinguishbetween objects. Successful development

of an imaging detector for site remediationwill provide useful baseline information fordesign of a battlefield-deployable land orshoreline mine imaging system.

DEMONSTRATION SYSTEM DESIGN

Nonvolatile Elec&orrics’NVS5B 15 sensorswere employed for this project.~, 1994]NVE is at present the only commercialvendor of GMR sensors, afthough otherelectronics companies (for example,Honeywell and Motorola) arc expected tooffer GMR products in the next few yeara.GMR sensors detect a single vectorcomponent of an applied field, like Hallgenerators or pulsed inductive detectors, sothree orthogorrd sensors banks wifl be

3

GMR Sensor Array

Ferrousobject Readout Electronics

,m

[mcmmmm❑ mmmmclclm■ m==mimmm

Display

comers

1:

others

7

-mmmmmmmm❑ mmm■ ■ ■ ■

WElmmmFigure 3. Layout of the image acquisition apparatus.

necessary for full 3D imaging capability.A single 2D array of sensors was selectedfor tiIS demonstration where rdtematesensors have orthogonal axes of sensitivity.The checkerboard layout of the 5x5 array ofsensors is illustrated in F@re 2. In orderto simplify interpretation of the magneticimages at the end of the reporL the outputsfrom the sensors with vertical and horizontalaxes of sensitivity are displayed separately,as illustrated on the right side of thedrawing. The performance of the 5x5 sensorarray has beerr compared with a 3x3 array(not shown) where all the elements have thesame vertical axis of sensitivity. The 3x3-y uses the same printed circuit boardlayout as the 5x5 so that the effect ofvarying the sensor spacing by a factor of two

could be determined.

A schematic of the imaging system is shownin Figure 3. The sensor amays wereinterfaced with an electronics chassis thatcontained a 15V and 5V power supply.

The outputs of the sensors were connectedto a National Instruments 64-charrrreldataacquisition card which was insralled in aPentium PC. National Instruments’LabView software was used to acquire theimages that follow. The images areunprocessed beyond resizing and adjustmentof the grayscale for printability. Sincethere are only 25 sensors per image, the datafiles are only 400 bytes (25 sensors x 16 bitsper sensor) in size. Each image is anaverage of 1000 readouts of the full arrayduring a 10-second period (acquisition rate= 100 Hz), although there is no reason thatdata could not be acquired much morerapidly (10-100 kHz). 1000 readouts of thearray was decidedly overkill; images werenot degraded by the averaging of smallerdata sets.

A variety of ferrous objects were imaged.These included tools, bolts, nails, rebar andpermanent magnets. AU objects wereimaged in their rcmanent magnetic state

4

(i.e., no external applied field) except whereotherwise specifically noted. Before animage was acquired, the no-object output ofall the sensors was obtained using the PC.This background signal represents acombination of offsets in the sensors, thesensors’ response to the earth’s field (nomagnetic shielding was used) and theirresponse to magnetic objects in thelaborato~ where the data was acquired, e.g.rebar in the floor. This background signalwas saved to a file and then subtracted fromsubsequent data. Objects to be imagedwere placed typically 1.5 cm above thesensor array on a Iexan stand. The falloffof the signal from the may with separationwas studied by stacking fmbricks betweenthe array and the ferrous object. Largerobjects such as rebar could be detected at ameter separation (signal:background ratio of2:1) although there was no real image at thatseparation with the 12cm-square array usedfor this demonstration.

RESULTS

A sampling of images produced with theGMR sensor array is shown in Figures 4-7.Figure 4 shows an image of a #10 threadedrod 1.5 cm above the array, as pictured inthe top-view drawing on the right. The twogray-scale images on the left are dataobtained from the sensor array. The topimage shows data from the GMR elements(labelled “comers”) with a vertical axis ofsensitivity. In the “comers” image, a valueof Ovolts is displayed at the positionscorresponding to the elements with ahorizontal axis of sensitivity. The bottomimage, labelled “others,” shows data fromthe GMR elements with a horizontal axis ofsensitivity. Xnthe “others” image, thepositions corresponding to the “comem”elements are displayed as Ovolts.Comparison with Figure 1 will clari& whichpixels are meaningful in the two images. Ina real UXO detection system, moresophisticated software would combine thetwo images in a contour or vector plot.Here darker grays indicate higher magneticflux, while lighter grays indicate lowermagnetic flux. The scale is on the right ofeach image. The numbers next to the

grayscale are the sensor signal in volts, sothat 1.4e-2 means 14 mV of signal.

In the top image of Figure 4, the magneticpoles on the ends of the rod are bekg pickedup by the sensors at the upper right andlower left comers. In the bottom image,the sensors are responding to magnetic fluxleaking from the sides of the rod. While thecharacteristics of the rod are not completelyclear with this low spatial resolution, itsgeneral shape and size of the rod can readilybe determined.

Figure 5 shows another image of the samerod, only this time flipped over so that it ispointing towards the opposite comers of thearray. The movement of the magneticpoles to the upper left and lower rightcomers is obvious in both images. Thereare several reasons why this image is not aperfect mirror of Figure 4, namely differentlateral placement of the rod on the array anddifferent rotation of the rod about its ownaxis. The magnetic domains in the rod maynot be azimuthally symmetric, with theresult that the image may depend somewhaton which side of the rod is facingdownward. The rod in Figure 5 is aIsooriented differently with respect to theearth’s field than in Figure 4. In a real UXOsystem possible ambiguity created bydifferent remanent states of objects can beaddressed through application of a rotatingalternating-current magnetic field created bytwo orthogonal sets of coils. A field largeenough to force maegneticpoles on eachsurface of a permeable object will makeeach surface visible to the GMR array,which is in essence a magnetic edgedetector.

Figure 6 now shows an image of the samerod in a constant 6 Oe external applied fieldwhich was generated with air-coreHelmholtz coils. Before acquiring theimage, the background of the array wascharacterized in the presence of the 6 Oefield. The most striking part about thisimage is that it looks much like F@re 4,showing that the GMR sensor array is ableto image ferrous objects even in thepresence of a substantial backgroundmagnetic field (about fifteen times the

5

,.

Corners I

Others ~

I-1.4 EBE-2

-7.8 BEE-3

-9.t3BEE+E

I-1.4 EEE-2

-7.9 E!EE-3

-B. BODE+9

Rod above Sensor Array

❑ mmmm

■ ■ ■

‘H

■ ■ ■■ ■

■ ■ ■

Figure 4. Image of a threaded #10 rod placed above the GMR array. On the lefi the gray-scale images show the response of the elements to the magnetic field emanating from the rod.The “comers” image shows only data from the elements with a vertical axis of sensitivity, whilethe “others” image shows data from the elements with a horizontal axis of sensitivity.

ea& field). The reason for the similarity ofthe two images is that the army detectsspatial magnetic field variations, not thescalar magnetic field amplitude like acesium-vapor magnetometer. Figure 4supports the assertion that array-baseddetectors will usable with magnetic soils aslong as the ground clutter is reasonablyhomogeneous on the length scale of theobjects to be detected.

Another point about Figure 6 is that theimage is a bit clearer than in Figure 4, where

no external field is applied. The improvedimage quality occurs because much of theflux from the applied field passes throughthe magnetically soft rod. A more completeoutline of the rod could be made byacquiring another image with so appliedfield in the orthogonal in-plane direction.In fact, a real UXO system would likelyincorporate 3 sensor arrays, each with adifferent orthogonal axis of sensitivity.Data would be read out from each arraywhile a coil applying a magnetic field alongthat direction is energized. A fully realizedsystem would incorporate a rotating

6

magnetic field and synchronous acquisitionfrom the 3 orthogonal arrays. Anadditional group of 3 sensors could be usedwith a portable GMR detector to eliminatenoise due to motion of the detector in theearth’s magnetic field. (Such a noise-elirnination scheme has recently beendescribed for a fluxgate vectormagnetometer system by Allen et al. [Allen,1996])

Figure 7 illustrates the ability of the array toimage slightly more complex objects. Heretwo bolts have been placed 1.5 cm above thearray. The “comers;’ image shows

Corners!

Others ~

substantial flux from the threaded part of thebolt, while the “others” image shows a moredltlcult to interpret pattern of flux possiblyarising from complex domain patterns in thebolt head. The overall “V” symmetry ofthe objects is apparent in both images. Theimage of the two bolts would be greatlyimproved by application of a rotatingexternal field and by a higher resolutionarray, with more pixels on each object.

I-5. EEEE-3

-2.5 EOE-3

-0.008E+0

I-l.l EflE-2

-5.5EOE-3

-fMEIEE+E

Rod above Sensor Array

m

■ ■■ m

■ ■ ■■ ■ ■■ mm

Figure 5. Same as Figure 4, but with the rod flirroed about a vertical axis. The image is almost. .a r&ror reversal of F;gure 4.

7

Another study was done to follow theevolution of images as a magnetized object(here a length of 3/4” rebar rod) is movedaway from the array. When the rebar was

Corners 1

Others ———+

images at different separation (not shown)demonstrate that we have an object with averticaf axis of symmetry. The imageformed by the sensors with their axis of

I-2.10aE-2

-1.050E-2

-0.000E+O

I-1.2EiOE-2

-6.E100E-3

-0.000E+O

Rod above Sensor Array

■Blr■ ■ ■■ ■

4 +

6 Oe external applied field

Figure 6. Same as F@re 4, but with a 6 Oe external appficd field acting on the rod and thearray. The image looks similar to F@rre 4 and is even a bh clearer despite the necessarysubtraction of the background signal from the external field. llris image suggests that GMRarrays will be able to locate ferrous objects even in magnetic soils of volcanic origin.

close to the array, it blinded the detector,saturating most of the sensors. As the rebarwas moved from 1.5 cm to 9 cm separation,the signal level was reduced from 280 mV to210 mV, still well above the typicalbackground level of 20-50 mV. The two

sensitivity paraflel to the rebm axis has thesame qualitative features independent ofspacing. In contrast, the image with theorthogonal axis of sensitivity variesdramaticaffy as a function of spacing. Thisvariation is due to different spatial falloff of

8

Corners I

Others ~

1-5.6 EBE-2

-2.8 EIEE-2Two

_-. -f813EE+0+O

bolts above Sensor Array

I-5. EEOE-2

-2.5 E9E-2

— -E. E8t3E+9

Figure 7. Image of two ferrous bolts placed above the array. The outline of the bolts is notdirectly visible, but the symmetry of the pattern is recomizable. More GMR elements andconco~tant higher spati-d resol~tion co~ld substantial~y improve this image.

the various muhipole components of the analysis of pulsed elechomagneticmagnetization pattern. One must keep in induction data, for example.mind that the magnetic field emanating froman object cart vary in all three dimensions, REALIZATION OF A FIELDABLE UXOand there is no particular reason in the DETECTION SYSTEMabsence of an external applied field for thesymmetry of a 2D slice taken at one height There are sever-alobvious improvementsto be exactly the same as a 2D slice taken at that would be necessary for a real-worldsnother height. On the other hand, the UXO detector. For example, there areimages sometimes appear rather simple, as questions about portability and ruggednessin Figure 4. Intelligent synthesis of data and of a fieldable GMR array system. In thisinterpretation of images will be the major regard it is worth noting that the powerchallenge in building a useful GMR-based consumption per GMR sensor (about 5 mWUXO detector, although the intrinsic dc for the NVS5B 15) is quite reasonable.difficulty is not greater than in time-domain This amount of powe can be provided by a

battery pack in a portable unit.

9

It should be clear from examination of theimages that having a larger array withadditional sensors will produce a moreimmediately recognizable result. There areno serious practical problems withconstructing a larger array. Ideally theindividual elements of a large array wouldbe addressable via row and columntransistors, much like a random-accessmemory or charge-coupled device array.Since UXO objects tend to be manycentimeters in extent, the GMR elements inthe proposed detector can be spaced farenough a~art that there is plenty of printed-. ..,,circuit b“oardarea available for these otherelectronic components.

For this demonstration, no signalconditioning electronics were employe~ thesensors are wired directly to the dataacquision card. A portable system withintegrated field-producing coils and 3-axissensitivity will require considerably moresophisticated signal conditioning andprocessing electronics. Since signal levels,data rates, and data amounts are all moderatefor this application, design of the supportelectronics for a GMR detector should bestraightforward. The implementation ofUXO-recognition software is moreambitious since magnetic pattern recognitionfor extended objects is still a new field.

Finally it is worth noting that NVE’s NVSsensors are the very fmt GMR-basedproducts to be commercially available;GMR was only discovered in 1988. Moresensitive GMR elements are expected to beavailable commercially later this year, withsubstantial improvement in sensorperformance expected in the near future.

Part of this work was performed underthe auspices of the U.S. Department ofEnergy by Lawrence Livermore NationalLaborato~ under contract number W-7405-ENG-48. Thanks are due to JohnAnderson and Russell Beech of WE andGary Johnson and Alan Wiltse of LLNL forassistance with this project. Financialsupport was provided by the ProgramDevelopment Office of the Chemistry andMaterials Science Department at LLNL.

REFERENCES

Allen, G., R. Koch and G. Keefe, UXOForurn ’96 Conference Proceedings,Williamsburg, VA, p. 1.

Anthony, T.C., J. A. Brug, and S. Zhang,“Magnetoresistance of Symmetric SpinValve Structures,” IEEE Trans. Magn. 30,3819 (1994).

Baibich, M.N. et aL, “GiantMagnetoresistaru of (001) Fe/(001) CrMagnetic Superlattices,” Phys. Rev. Mt.61,2472 (1988).

Binasch, G., P. Griinberg, F. Saurenbach andW. Zinn, “Enhanced magnetoresistance inlayered magnetic structures withantiferromagnetic interlayer exchange,”Phys. Rev. B 39,4828 (1989).

Chaiken, A., G.A. Prinz, and J.J. Krebs,“Magnetotransport Study ofFe-Cr-Fe Sandwiches Grown on ZuSe,” J.Appl. Phys. 67,4892 (1990).

Chaiken, A., P. Lubitz, JJ. Krebs, G.A.Prinz and MZ. Harford, “Spin-ValveMagnetoresistance Of Uncoupled Fe-Cu-CoSandwiches,” J. Appl. Phys. 70,5864(1991).

Kane, H., et aL, “Substrate TemperatureEffect on Giant Magnetoresistance ofSputtered Co/Cu Muh.ilayers,” Appl. Phys.Lett. 63,2839 (1993).

McGlone, T.D., Proc. Symp. Applications ofGeophysics and Engineering toEnvironmental Problems, Reno, NV, 1997,p. 705.

Nonvolatile Electronics (NVE) Inc., 11409Valley View Road, Eden Prairie, MN55344-3617, (800) 467-7141,http://www.nve.com, info@ nve.com.

10

Technical Inform

ation Departm

ent • Lawrence Liverm

ore National Laboratory

University of C

alifornia • Livermore, C

alifornia 94551