Applicationsof Ultrasoundto MaterialsChemistry

Kenneth S. Suslick

The following is an edited transcription ofthe address that Kenneth S. Suslick, recipientof an MRS Medal Award, gave at the 1994MRS Fall Meeting. Suslick received thishonor for "incisive studies of the chemicaleffects of ultrasound on solids and surfacesand the use of sonochemistry as a new syn-thetic approach to unusual, inorganic com-pounds or materials."

This article will begin with an intro-duction to acoustic cavitation, the physi-cal phenomenon responsible for thechemical effects of ultrasound. Some re-cent applications of sonochemistry to thesynthesis of nanophase and amorphousmetals, as well as to heterogenous cataly-sis, will then be highlighted. Finally, wewill examine the effects of ultrasound onmetal powders in liquid-solid slurries.

CavitationThe chemical effects of ultrasound do

not come from a direct interaction ofsound with molecular species. Ultra-sound has frequencies from around15 kilohertz to tens of megahertz. In liq-uids, this means wavelengths from cen-timeters down to microns, which are notmolecular dimensions. Instead, whensound passes through a liquid, the for-mation, growth, and implosive collapseof bubbles can occur, as depicted in Fig-ure 1. This process is called acousticcavitation.

More specifically, sound passingthrough a liquid consists of expansionwaves and compression waves. As soundpasses through a liquid, if the expansionwave is intense enough (that is, if thesound is loud enough), it can pull the liq-uid apart and form a bubble (a cavity).The compression wave comes along and

compresses this cavity, then another ex-pansion wave re-expands it. So we havean oscillating bubble going back andforth, say, 20,000 times a second.

As this bubble oscillates, it growsthrough several mechanisms, one ofwhich is rectified diffusion. In rectifieddiffusion, the surface area on expansionis slightly larger than on recompression,so growing processes are kineticallyslightly faster than shrinking processes.This oscillating, growing bubble reachesa resonant size determined by the fre-quency of the sound field. When the

Compression

bubble is in resonance, it is well-coupledto the sound field, it can absorb energyefficiently, and it can grow rapidly in asingle cycle. Once it has grown, however,it is no longer well-coupled to the soundfield. At this point, the surface tension ofthe liquid combined with the next com-pression wave implosively collapse thebubble on a submicrosecond time frame.A shock wave can be generated in thegas of the bubble in addition to thesimple compressional heating of the gas.When gas is compressed, heating results.When gas is compressed this rapidly, theheating is nearly adiabatic. The heat hasno time to flow out, so a very localized,transient hot spot forms, and that hotspot is responsible for the chemistry thatis observed.

The conditions formed during thattransient cavitation are extreme. We havebeen able to measure temperatures andpressures by comparative rate ther-mometry and by using sonolumines-cence as a spectroscopic probe of thespecies formed during cavitation. Ourcurrent best estimates of the hot-spotconditions give temperatures above5000 K, pressure of about 1700 atm, andtime duration under 100 ns, and the timemay be substantially less than that. Wetherefore have cooling rates associatedwith this process of more than 101" de-grees/s. For calibration purposes, if Ithrust a poker of red-hot iron into icewater, I get a cooling rate of a few thou-

Expansion

.52

CD

150 - |

100-

5 0 -

IMPLOSION

SHOCKWAVE

FORMATION RAPIDQUENCHING

100 200 300 400 500

Time . sec)

Figure 1. Transient cavitation.

MRS BULLETIN/APRIL 1995 29

Applications of Ultrasound to Materials Chemistry

Figure 2. Islands of chemistry as a function of time, pressure, and energy. Adaptedfrom "The Chemical Effects of Ultrasound," by Kenneth S. Suslick. Copyright©1989 by Scientific American Inc. All rights reserved.

sand degrees per second. If I splattermolten metal onto a liquid-nitrogen-cooled surface, I get cooling rates of a fewmillion degrees per second. We will re-turn to the implications of these coolingrates later.

This understanding of cavitation al-lows us to compare sonochemistry withother forms of chemistry. Fundamen-tally, chemistry is the interaction of en-ergy and matter. The parameters thatcontrol that interaction are the time ofthe interaction, the amount of energy inthe interaction, and the pressure, whichtogether describe the three-dimensionalspace depicted in Figure 2. This figureshows the heavily overpopulated islandof thermal chemistry at medium pres-sure, time, and energy. For high-pres-sure, long-time scales such as occurunder geological conditions, the graphshows the spiked island of piezochem-istry. The island of sonochemistry isnear photochemistry and flame chemis-try. All of these are related because theyare all forms of interacting energy andmatter. However, each has its ownspecific characteristics because each oc-cupies a different region of this three-dimensional space.

To introduce ultrasound into solutionsin the laboratory, we use a high-intensityultrasonic horn that consists of a solid ti-tanium rod connected to a piezoelectricceramic and a 20 kHz, 500 V power sup-

ply (Figure 3). This commercially avail-able apparatus can be thermostated andthe atmosphere above the solution can becontrolled. It is useful for small-scalework, which is mostly what we do. Notethat large-scale processing of liquidswith ultrasound also exists. Large-scalecleaning baths are available, and areused, for example, by the military, toclean intact jet engines. Flow reactorsalso exist and are commercially availablein stackable 20 kW units. The largest

Power Supply

Stainless SteelCollar & O-Rings

Figure 3. Sonochemical apparatus.

scale application of ultrasound that Iknow for the physical processing of aliquid was for coal benefication at20 tons/h.

Applications

The Synthesis of Amorphous andNanoscale Materials

Given the unusual conditions createdduring cavitation, we considered variouspotential applications, one of which wasthe possibility of using ultrasound as away of generating amorphous andnanoscale materials. Amorphous metalshave unusual magnetic, electronic, andcatalytic properties. To form an amor-phous metal we need high cooling rates(above a million degrees per second) sothat the material can be frozen beforeit crystallizes. Thermal quenching ofmolten metal usually requires the addi-tion of nonmetal alloying components;boron is common. Consequently, mak-ing pure amorphous iron has provendifficult.

We realized that given these cavita-tional hot spots, sonochemistry providesenormous cooling rates fast enough tocause solidification before crystallizationcan occur. However, the primary reac-tion site of cavitation is the gas phase in-side the bubble, in which case, we need away of producing metal inside the cavita-tion event. Drawing on the same ideasused in organometallic chemical vapordeposition (CVD), we need a volatile pre-cursor. We initially considered metal car-bonyls and metal nitrosyls. When theyare irradiated with ultrasound, we areable to generate metals from iron and co-balt complexes. From the early transitionmetals we tend to form metal carbides.

We used sonoluminescence as a spec-troscopic probe to see if, in fact, we canstrip the ligands off of our precursors. Ifwe start with iron pentacarbonyl, and ir-radiate with ultrasound in, say, a dode-cane solution, the light coming out isemission from excited-state iron atoms.This shows that iron pentacarbonyl is be-ing sonochemically decomposed to ironatoms, and that some of those iron atomsare in electronically excited states. Simi-lar atomic emission is observed fromother volatile organometallics.

Once we have these volatile precursorsin the cavitation event and we strip offthe ligands, we can form small clustersof metal atoms in that cavitation bubbleand can work with them. As shown inFigure 4, we can let them agglomerate toform amorphous metals. We can trapthem with a polymeric ligand, such aspolyvinylpyrrol idone and form a

30 MRS BULLETIN/APRIL 1995

Applications of Ultrasound to Materials Chemistry

n = 10- 1000

Figure 4. The sonochemical synthesis of amorphous metals, nanocolloids, andsupported catalysts.

nanocolloid; or we can deposit thesesmall clusters on an oxide support, suchas silica, and form supported heteroge-nous catalysts.

In fact we have been able to do allthree of these. If we irradiate iron pen-tacarbonyl carbonyl with ultrasound, ina relatively unreactive alkane, a highlyreactive, black powder is formed. It has ahigh surface area of about 150 m2/g, andit burns spontaneously in the air becauseof that high surface area. Under modestmagnification, the material is clearly notcrystalline and shows conchoidal frac-tures. It is relatively pure iron by elemen-tal analysis. On higher magnification,the material shows a porous, coral-likestructure (see Figure 5). On still highermagnification, the individual particlesthat make up this agglomerate are appar-ent, and nanometer-sized clusters can beseen that form the building block of thiscoral-like material. Initially as prepared,the material is amorphous by x-ray dif-fraction, by neutron diffraction, and bye-beam microdiffraction. Furthermore, itshows a one-time irreversible crystalliza-tion exotherm in the differential scan-ning calorimetry (DSC) at about 350°C.

As an example, characterization by x-ray diffraction (XRD) pattern shows nopeaks for the material as formed. As weheat the material to above 200°C, crystal-lization begins to set in, and by about350°C, the material is fully crystallizedinto normal a-iron. We have no evidencefrom XRD or other techniques for the

formation of carbidic or oxide phases.If we want to trap the material in the

small nanometer-cluster size, we can usea weakly coordinating ligand, such aspoly vinylpyrrolidone. To do that, we can

Figure 5. Porous, coral-like structureof sonochemically preparedamorphous iron. From K. Suslick,Ultrasonics 30 (1992) p. 171, bypermission of the publishers,Butterworth-Heinemann Ltd. ©.

add the polymer to the iron pentacar-bonyl solution and irradiate at room tem-perature with high-intensity ultrasound.If we want to make a supported catalyst,we can do the same thing, except wewould replace the polymer with an oxidesupport. We have characterized theseextensively. Initially, as formed, thecolloids are also amorphous. Electronmicrodiffraction shows no crystallinityassociated with them, and they undergoa one-time irreversible crystallizationexotherm in the DSC.

These very soft magnetic materialsshow virtually no hysteresis in magne-tization curves. In magnetic-propertystudies done in conjunction with MyronSalamon, a professor of physics at UIUC,the nanophase amorphous iron in bulk isa very soft ferromagnet. Its magnetic mo-ment lies between crystalline iron andmolten iron. We have agreement in mag-netic measurement from SQUID andneutron-diffraction data. The effectiveexchange is relatively modest betweenthe irons — about 30% of crystallineiron—and this has been modeled with acorrelated spin-glass random packingmodel.

The nanocolloidal iron is superpara-magnetic (that is, essentially a single do-main ferromagnet that can be thermallyoriented). The nanophase iron supportedon silica is also superparamagnetic, withproperties similar to those of the colloid-al iron.

Catalytic PropertiesAmorphous surfaces are interesting

catalysts for many reasons. They havehigh concentrations of low coordinationsites; that is, they are heavily defected.They are roughly single phase, so weneed not be much concerned aboutwhether {111} or (100) planes, for ex-ample, are exposed. In other words, theyare relatively isotropic surfaces. Theproblem with amorphous metals as cata-lysts is they are very difficult to makeand usually have low surface areas. Ingeneral, the surfaces are also heavilypassivated during processing. The otherdisadvantage is that the materials are notin an equilibrium state and are likely tocrystallize.

To examine the catalytic reactivity ofour amorphous powders, we use an in-dustrial-strength catalytic microreactor,which is basically a glorified digital gasmixer that flows appropriate gasesthrough a catalyst bed, consisting of ourpowder, and into a gas chromatographmass spectrometer (GCMS). The reac-tions we initially chose had a bifurcatedpathway; that is, two products were pos-

MRS BULLETIN/APRIL 1995 31

Applications of Ultrasound to Materials Chemistry

sible. Thus, we could look at selectivitydifferences rather than argue aboutwhich catalyst was more active. For ex-ample, the reactions of cyclohexane overmetals can lead to either formation ofbenzene—a highly desirable process—or the hydrogenolysis to methane—anundesirable process. My graduate stu-

dent, Taeghwan Hyeon, recently discov-ered that we could use our amorphousmetals and a series of alloys to dramati-cally influence the selectivity (Figure 6).This came as a complete surprise. As isknown in the literature, iron and cobaltare bad catalysts for the dehydrogenationof alkanes. They do not lead to the for-

82

17

100 i

CDCCDN

CDDQ

250 275

Temperature (°C)

300

Figure 6. Catalytic activity and selectivity of nanophase Fe-Co.

mation of benzene, but rather to crackingto produce methane. Our amorphousiron and our amorphous cobalt bothturn out to be poor catalysts for dehydro-genation. But the alloys are superb. Wecannot yet account for the origin of thisphenomenon.

We can make supported catalysts, as Imentioned earlier. For example, we havesonochemically prepared silica gel withdeposits of iron nanometer clusters. Theiron clusters cannot penetrate into thesilica, so we get essentially an eggshellcatalyst. This differs greatly from the re-sult of normal methods where, for ex-ample, we deposit an iron nitrate solutiononto silica. If we look at higher magnifi-cation on the TEM, we can see those fewnanometer-sized clusters against thegray amorphous silica background. Andagain, the clusters are a few nanometersin size in these materials.

These supported catalysts are ex-tremely active. In this case we are lookingat Fischer Tropsch synthesis hydrogena-tion of CO, to form low molecular weighthydrocarbons. Both conventional andsonochemical methods were used to pre-pare catalysts. At low temperatures, thesonochemically prepared catalyst ismuch more active than the conventionalcatalyst with similar concentrations ofiron and similar dispersions. As we heatthe sonochemically prepared catalyst tothe point of crystallization, we begin tolose much of the activity, leading usto suspect that this increase in activitycorresponds to the high defect concen-tration in these nanometer-scale clusters.

Heterogenous SonochemistryI will now touch briefly on ultrasound

applications that involve liquid-solid re-actions. Because of space confinement Iwill simply mention that there are nowhundreds of examples of the use of ultra-sound to drive liquid-solid reactions, es-pecially of highly reactive metals such asLi or Mg. When we began, we decided tofollow the Zeroth Law of Engineering,which is "If it works, don't fix it." So wewent looking for the world's worst het-erogenous catalyst.

We decided that near the bottom of thelist is nickel powder right out of thebottle. Raney nickel (a porous form of themetal) is, of course, a very active hydro-genation catalyst but it is also expensive,pyrophoric, and environmentally prob-lematic. The point is, though, that if wetake ordinary nickel powder and irradi-ate it with ultrasound, we can increaseits reactivity by more than 100,000-foldand regain Raney nickel-like activity.

When we study the morphology of the

32 MRS BULLETIN/APRIL 1995

Applications of Ultrasound to Materials Chemistry

powder, we find that before sonicationthe nickel powder is crystalline on thesurface. After ultrasonic irradiation, thecrystallites are removed and the surfaceis smoothed on a macroscopic scale(which means that it is roughened on anatomic scale), as shown in Figure 7.

If we examine sonicated nickel powderat lower magnification, we see agglomer-ation occurring as well. It looks like ballmilling on a microscopic scale. If the par-ticles hit at a glancing angle, material isbrushed off and smoothing occurs; if theparticles hit directly, they stick. The con-sequence is that we remove passivatingcoatings on the surface. Auger depthprofiling reveals that, before ultrasonicirradiation, a deep oxide coating ispresent. That is why these metals are un-reactive. If we irradiate with ultrasound,that oxide coating is removed.

I would like to close by examiningmore closely the effects of ultrasound onslurries. Figure 8 shows where particleshave collided and generated what ap-pears to be a melted neck between them;If we do this with two different metals,we can actually do elemental Augermapping and demonstrate that the neckis an alloy between the two differentmetals—tin and iron. For example let usassume that the spot welding on collisionrepresents true melting (although wemay debate whether it is fully melted orat least partly a plastic deformation). Themelted neck size is about a cubic micron.About 90 KJ/mol is needed to melt ametal such as iron. That means 0.1 ergwas needed to melt the neck. We cantake that energy to be a lower limit forthe kinetic energy of impact between theparticles when they collided. This givesus a velocity on the order of hundreds ofmeters per second, a significant fractionof the speed of sound in these liquids.That is impressive: We have a flask of acold liquid, an ultrasonic horn in it, andwe generate interparticle collisions be-tween particles at velocities that are halfthe speed of sound.

The origin of these high-velocity colli-sions comes from acoustic cavitation. Thecollapse of bubbles during cavitation islike setting off a microscopic depth-charge in the liquid, and it generatesshock waves that travel through the liq-uid at or above the speed of sound.When such a shock wave comes acrosssolid particles in close proximity, it canaccelerate one particle and smash intothe next. If the particles hit at a glancingblow, we get smoothing; if they hit di-rectly, we get spot welding at the pointof impact.

What kind of local conditions occur

Before U.S.-160 microns

60 min. U.S.-80 microns

Figure 7. Surface Morphology of Nipowder (a) before ultrasound showingcrystallites on the surface and (b)after ultrasound, which removesmaterial and smooths themacroscopic surface. FromK. Suslick, Solid State Ionics 32-33(1989) p. 447.

Figure 8. Neck formed by particles ofzinc colliding in a slurry irradiatedwith ultrasound. Reprinted withpermission from K. Suslick, Science247 (1990) p. 1067. Copyright 1990American Association for theAdvancement of Science.

when these particles collide? We canprobe these conditions by examining aseries of metal-powder slurries. For ex-ample, we can look at chromium, molyb-denum, and tungsten with the sameparticle size, concentration, inert sol-vents, and ultrasonic intensity. For chro-mium we see tremendous agglomerationbefore and after ultrasonic irradiation,and at higher magnification, we see sub-stantial deformation of the individualparticles in the scanning electron micro-graphs (Figure 9). Chromium melts at1800 K. Molybdenum melts at 2600 K.Agglomeration still occurs, but it is notnearly as dense; and with higher magni-fication in the SEM, no smoothing or de-formation of the individual particle isobserved. Tungsten melts at 3400 K, andthe micrographs show that ultrasonic ir-radiation has no effect.

If we make a chart of different metals,with their corresponding melting points,occurrence of agglomeration, or changesin surface morphology, we discover abreaking point around 3000 K. That tem-perature is unrelated to the temperatureof the hot spot, but it is another indica-tion of the extreme conditions that canbe formed in liquids irradiated withultrasound.

ConclusionThrough the process of cavitation, ul-

trasound performs high-energy chemis-try. There are applications of this to thesynthesis of inorganic materials, amor-phous metals, alloys, nanophase colloids,and supported catalysts. There are otherbiomaterial applications through thesynthesis of protein microspheres. Theseprotein microspheres have applicationsfor medical imaging, drug delivery, andblood substitutes, but are beyond thescope of this discussion. The applicationsof high-intensity ultrasound to materialsscience are diverse and the range ofpossibilities is only now beginning toemerge.

AcknowledgmentsI would like to thank all of my stu-

dents and co-workers who have been in-volved in this project over the years: forthe work discussed in this article, expe-cially Dom Casadonte, Seok-Burm Choe,Andrzej Cichowlas, Steve Doktycz, MingFang, Mark Grinstaff, and TaeghwanHyeon. I thank Prof. Myron Salamon forcollaboration on magnetic properties,and Profs. Galli and Bellisent for collabo-ration on neutron diffraction. This re-search has been funded by the NationalScience Foundation and the Exxon Foun-dation, and has been further supported

MRS BULLETIN/APRIL 1995 33

Applications of Ultrasound to Materials Chemistry

Before Ultrasound After Ultrasound

Chromium (m.p. 1857°)

Molybdenum (m.p. 2617°)

Tungsten (m.p. 3410°)

by UIUC Materials Research Laboratoryand the Center for Microanalysis of Ma-terials, which is funded by the Depart-ment of Energy.

Some Leading References toMaterials Science Applicationsof SonochemistryK.S. Suslick, Science 247 (1990) p. 1439.K.S. Suslick, in Encyclopedia of Materials Sci-

ence and Engineering, 3rd suppl., R.W.Cahn, ed. (Pergamon Press, Oxford, 1993)p. 2093.

M.W. Grinstaff, M.B. Salamon, and K.S. Sus-lick, Phys. Rev. B 48 (1993) p. 269.

R. Bellissent, G. Galli, M.W. Grinstaff, P.Migliardo, and K.S. Suslick, Phys. Rev. B 48(1993) p. 15797.

K.S. Suslick, and M.W. Grinstaff, in Macro-molecular Assemblies, P. Stroeve and A.C.Balazs, eds. (Am. Chem. Soc, Washington,D.C., 1992) p. 218.

R. Roy, /. Sol. St. Chem. Il l (1994) p. 11.

Kenneth S. Suslick, recipient of the 1994MRS Medal, received his BS degree from theCalifornia Institute of Technology and hisPhD degree from Stanford University. Hejoined the University of Illinois at Urbana-Champaign in 1978 and now holds a jointappointment there in the Departments ofChemistry and of Materials Science andEngineering. Last year, Suslick received theAmerican Chemical Society Nobel LaureateSignature Award for Graduate Educationtogether with his former graduate studentMark Grinstaff. He is a Fellow of theAmerican Association for the Advancement ofScience, and has received a National ScienceFoundation Special Creativity Award, a SloanFoundation Fellowship, and a NationalInstitutes of Health Research Career Develop-ment Award. For more information, Suslickcan be contacted at the following address:

School of Chemical SciencesUniversity of Illinois at Urbana-

Champaign505 S. Matthews AvenueUrbana, IL 61801phone: (217) 333-2794fax:(217)333-2685e-mail: ksuslick@uiuc.edu.

Figure 9. Agglomeration of metal powders after ultrasonic irradiation of decaneslurries. Reprinted with permission from K. Suslick, Science 247 (1990) p. 1068.Copyright 1990 American Association for the Advancement of Science.

Have you seenMRS Homepageon World Wide Web?http://dns.mrs.org/

34 MRS BULLETIN/APRIL 1995