I. CharacteristicsII. Theoretical and Experimental Results

III. Fragmentation MechanismsIV. Field ObservationsV. Summary and Future Research

GLOSSARY

phreatomagmatism Volcanic activity resulting from the in­teraction between magma/lava and groundwater or sur­face water including seawater, meteoric water, hydro­thermal water, or lake water.

Fel An abbreviation for fuel-coolant interaction, referringto an industrial analog to phreatomagmatic explosioninvolving a fuel such as molten iron and a coolant suchas water. The interaction of dle fuel and coolam hasresulted in vapor explosions in indusu'ial environmentsand are a potential hazard associated with nudear coremeltdown.

superheated liqUid A metastable thermodynamic state of aliquid resulting 6:om rapid heating to a temperature wellabove the boiling point.

Enryclop.dia of Voh'mweJ 431

" MEGHiAN MORRISSEYColftl'ndo Schvol OfMi,le.,\'

:BERND ZlMANOWSKIUnl'uemt:y of Wttef:z;hurg

Kr:,.NNETH WOHLEI Z'£Q)' AI1l1/lQS NatiOtlti/ Laboratory

&AlL BURn" ;RUnlvfl'sity of WU810zburg

Kelvin-Helmholtz and Rayleigh-Taylor instabilities Interfa­cial fluid instabilities caused by relative motion of tV{Qinuniscible fluids. Kelvin-Helmholtz i.nstabiJi.ties are in­duced by shear stresses along the interface, whereas Ray­leigh-Taylor instabilities develop during accelerationof a denser fluid by a less dense fluid. In both casesfragmentation occurs when surface tension forces are ex­ceeded.

vapor film A layer ofvapor formed at dle i.nterface ofwaterand a bot liquid or solid body (temperature ~ boilingpoint).

wet/dry eruptions At standard conditions (25°C and 0.1MPa), water is a two-phase (liquid and vapor) fluid. De­pending on the quantity of available heat energy, whenwater comes into contact with magTIla it may become aliquid-rich fluid or a vapor-ricll fluid, resulting in a dryeruption or a wet erupti.on, respectively.

IT WAS FASCINATING to watcb tbe lava C1'llterand glowing rivers and rivuletsflowing down tbe slopesofthe lava dome. Stillm01-e exciting, however, was the-relentless struggle bet"JJeen the advancing lava and thesea. From the very beginning of the lava eruption itwas noteworthy how much of the lava became frag­mented when it came into contact with the sea.

-s. Thorarinsson, 1966

Copy"ighr © 2000 by Ac:1demic PressAll righrs of reproduction in ony forlll reserved.

432 PH TO~IA M Ie FRAG 1 An0"'"

Th.e ab e quote .. as made by Thora.riD 'on as h b­served the eruption of urtuI volcano and the formationof a new island ( urtsey) off the saud] central coast ofIceland in 1964. His observations of this eruption ledto a new cia smcation f v lcanisll1 1m wn as phreato­magmatism and dle awareness of fragmentation pro­ces es a iated with dl interaction ben een magmaand n nm gmatic ater.

I. CHARACTERISTICS

A. Global Dccm-rence

Phreatomagmatic aetivi involves dle ph ieal interac­tion of magma or lava \vim an eternal ouree of at I'

either widUn the Earth r on its urrnc. spectrum ofaCtlVl r when magm.a cont:3c water as illustratedby the foil \'ling exampl . (1) ince J990 Ja a ha b enpouring into the sea al n the shor near KalapanaHawaii producing a range of activity &om the explo'iveejection f fragmented lava to a continuous emission ofteam and passive quenchin.g· of lava (Fig. la). (2) The

subgla ia1 mption a rimSVOtll v leano celaud) i.n1996 ca ed exteusi e flo ding in ouch central Icelandand pr duced explosive je of a b (Fig. lb). (3)Moun Ruapehu e, ~Jand) in 19 ,a vigorouprush water. Yap r and oria bomb was ob erv

through the crater lake producing c ck's tail expl ivejets fa h. (4) In 1977 the two Ukinrek aars ( a k:J)formed by several hundred periodic blasts which wereaccompanied by a pervasive production of steam whenri ing magma surfaced into a shallow lake.

The aHa of eruption styles as oated \'lith phreato­magma .c activi. r Beets the man environmen inwhich magma or la a may encoon er water. peciJicenvironments include eep or hallow submarine li o­ral lacuscr:ine phreatic and subglacial Phreatomagma­tism' nOtrestricted to magma typ or em type (mono­genetic or polygenetic) and reBec the pervasi eoccurrence of water in the crust. olcanoes that buildup fi'om repeated eruptions (polygenetic) such as strato­volcanoes and shield volcanoes, oft n have a plu·eato­magmatic phase or phases (i.e., vuleanian, phreatoplin­ian or surtseyan) during an eruption period. Thedistinguishing featur of a phreatomagmatic eruptionpha are that plum end to be more earn or \ a errich and the grain .ze of the eruptiv produc tend to befin r than those from the a ociated magmatic erupa epitas . Explosive phreatomagmaac activity is al 0 cllar­actel1zed by intensive fragmentation an.d ejection f thewall r cks in dle vicinity of dle xplosion site, whichre ult in the formation oflarge craters. Volcan s dlatbilll up from a ingJe phreatomagmatic eruption (mon­ogenetic) form either a tuff ring, cuff cone or a maarolcano when rising magma CODt:3C water. The depo ­

its of cll volcano rna cons:' f orne 10 morethan 1 0 ingle la e and r uJ dominand from

FI G U REI (A) Photo of the lava from Kilauea volcano entering the coast near KaJapana In 1992 (taken by Thorvaldur Thordarson).(B) Photo of the October 2. 1996 eruption column after it broke through the ice at GrimsvoUl volcano (COUrtesy of FreysteinnSigmundsson, Nordic Volcanological Institute).

PHREATOMAG,\lATIC FRAGMENTATION 433

transport and deposition by pyroclastic surges. Thesetypes of volcanoes are found all around the world andinclude Craters of the Moon, United States; DeceptionIsland, Antarctica; Eifel volcanic field, Germany; IwoJima, Japan.

B. Eruption Styles

A phreatomagmatic eruption is driven primarily by thevolumetric expansion of external water after it has beenrapidly heated by contact widl magma. Exsolving mag­matic volatiles may also contribute to expansion andfragmentation during a phreatomagmatic event. Menwater comes into contact with magma, it will eithertransform to steam (vapor) or a two-phase (liquid andvapor) fluid depending on the relative masses of waterand magma interacting. It may also become superheatedto a metastable liquid state which will flash to steamwhen acted on by an external force. In general, the morewater available during the interaction, the wetter (liquid­rich fluid) the eruption will be. There is a wide varietyof eruption styles that results from magma/water inter­action. The style ofactivity ranges from passive quench­ing and granulation of magma or lava to large scalethennohydraulic explosioIl5, representing dle most en­ergetic result of magma/water interaction.

The most commonly recognized phreatomagmaticeruption style is dlat characterized by cypressoidal(cock's tail) explosive jets of ash accompanied by bil­lowing clouds of steam. This style of activity was welldescribed during dle emergence of Surtsey volcano offthe coast of Iceland in 1964 and appropriately namedsurtseyan. Ejecta are directed vertically and laterallywith the laterally directed ejecta producing a base surge.Similar activity has been observed at Capelinhos(Azores), Taal (Philippines), St. Augustine (AJaska),Ruapehu (New Zealand), and Vulcano (Italy) wheremagma has come into contact with either seawater orlake water filling a crater. Deposits related to this styleof activity have been described as wet deposits, indicat­ing dlat not all of the water was converted to steamduring eruption and deposition.

Other phreatomagmatic eruption styles include vulca­nian and phreatoplinian. Vulcanian eruptions are dis­crete, explosive events that last seconds to minutes.These eruptions are thought to be caused by highlypressurized gases derived from the magma or fTom aphreatomagmatic interaction. Phreatoplinian eruptionsare similar in style to pHnian eruptions except that de­posits are finer-grained. Deposits related to these two

styles of activity are dry relative to sUl·tseyan depositsin that most of dle water was converted to steam duringeruption and deposition.

C. Tephra Characteristics

The products from a phreatomagmatic eruption are pre­dominantly a mixture of glass, clystals, ~llld foreign orwall rock lithics (fragmented material fl"Om the conduitand vent). In many cases, the lithic fragments represcntby far the dominant part of the ejecta with grains rangingin size from fine ash to blocks. Their shapes may indicatefragmentation by fracturing (i.e., angular edges) and notby abrasion (i.e., rounded edges). In tertuS of grain size,phreatomagmatic tephra generally are finer grained thatlmagmatic tephra (Fig. 2a), particularly for high w3ter/magma ratios.

Under optical 3nd electron microscopes, the juvenile(glass) component of phreatomagm:ltic tephra containsdistinct grain shapes. These shapes include blocky, fusi~

form, mosslike, phltey, and spherical or dl"Oplike asshown in Fig. 2b. Other distinguishing textural fearuresare particles adhering to glass surfaces, grooves orscratches, chipped edges, and rounded edges. AJso, theglass constituents may be more hydrodlermally alteredthan magmatic tephras of equivalent ;lge, especially insamples from wet tephra deposits in which hot water wascaptured. Basaltic glasses alter to palagonite, whereasrhyolitic glasses alter to a fine-grained mosaic of quartz,potassium feldspar, and clays.

D. Phenomonology

The physical dynamics of phreatomagmatic fragmellCl­tion parallel a well-studied process called fuel-coolantinteraction (FCI). FCI refers to a heat transfer processwhich can occur in any environment, convertingthcnnalenergy into kinetic energy on a short time scale «1ms) during the interaction of a cold volatile fluid (cool­ant) and a hot fluid (fuel) with a solidus temperaturewell above the homogeneous nucleation temperature ofthe coolant. In volcanological terms, the fuel is themagma and the coolant is water. Water at atmosphericconditions has a homogeneous nucleation temperatureexceeding 300°C. Magmas have temperarures well abovethis temperarure, therefore, in principle, phreatol11ag­matic explosive eruptions can occur in any volcanic envi­ronment.

434 PHREi\TO!\tAGM I FRJ GMDIT TI

b

II. THEORETICAL AND

EXPERIMENTAL RESULTS

A. History

Fcr ma b xplo ive or non xplo ive depending onthe rate of heat tran.sfer bet\ e n the fuel and coolantand the ub quent pressurizati n rate of the coolan .The rate of heat transfer i a function of th surfacearea of the fu Lwhich, in an expl ive FCl, increa e ba cascading fragmentation proc s. he &agm uta . nprocess i enhanced by thermal xpansion of the 0 hlJltthat create new contact surf-ace .

Th physical dynamics of an explosive FCr are repr ­nted by dle folJowing stages:

1. Initial contact and coa mixing of fuel and co 1-ant under stable vapor .film boilin conditionsid n­frost phenom non).

2. Comple apor film collap e caused either b thepas age of 3 pr ore puIs which rna be of e;...: emalorigin (e.g. ei micity) or b :l local implosion (i.e.,water hammer) due to rapid conden ation of co lantapor. Once the coolant vapor i c rnpLetelyconden ed,

the fuel and c alam are dlemlally and mechanicallycoupled.

3. Episodic increase of heat trm fer from fuel to cool­ant and fine fragmentation leading to superheated andpressurized water. the coolant' heated, i expandat a rate of <1 ms leading to rapid increa e in I adstress on the melL Relaxation ofI ad tress in th brittlmode (ela tic rebound) cans the eX"J>losive relea fsei mic energy. p to 90% of til tOtal kinetic expl i nenergy is r 1 ed in this ph e.

4. olumetric expansion of the fuel-eoolant mh:rurefrom the cran formation of th up rheated ater to

superhe3ted team. t this stage dl fuel and coolanare dlennaUy and mechanically d c upled.

All FCls, explosive or nonexplosive, are initiat d byStage 1. A 11 nexplosive FCr terminates at Srag"e 1 r

rage 2. The wetness of a11 eA.-pJo ive Cl reflects thedegree of fragmentation and the rate of heat transfer(or the efficien of the feedback mechanism) during

rages 2 and 3.

PLANAR lBASAL) BEDFORMS

l'IISSIVE BEDFORMS

11;1. 3 ;,~-.,..:::.~. .' .. ...,

:-~(~

•.':-2-.+,---±O.......,'~,...2 ---:-3~4--...,1~'!-'11I.IIIIt l2JQ ll,aq fUQ fD..25l 1/1.1~ l~ ffl~ la.-

Md.lmml!it.~...-.- ~lttal.. 1,.tAhpoll'. - w~ fItS80"", _,_

SANDWAVE BEDFORMS

~.c.:~.

..... .'...:,........,....

o.2~.~'---!-.---:!'--""'!---:!-3--!-4---!:1~6:-'tfml t2D:a (1. fQ.1q FUDf lI1'iI'I ra,DIJ fD',=t ~

Md.,tnm)SL "-1.,.. _.- i(unu. tin,st. ........ -- HobUlu, ,'',__ - F1'UHtWi.... 1113_.... -_t_,-

a

1 •

",·3 '/~-"~\.'2 ........-f., ....• ~.

o...':-..&..,~a-,'--:!-, ---:-3---L4---":--6':-" 0., ., 0 I , 3 4 I 6('JIOII (2..CIq (I.o:Jl l,1I,lIQ\i lUlIJ tl11:JJ {O.oll to.o:n (OD2] 14..o11't lltCOl (l,OlPj fO,lIQj 10.::::') 10..1.31 Io.~ IItIAQ CO.DCl)

Md.(mm) Md.lmm)SL HtlIl.n .. fAah CIQUI:Sl -- K~f1'l.t 1iI1.• 11UI1 ~H, Htlllllll -~- Hoblit! 1!!I1 01'1 '111151. H.'RIlI lAc:c:retlOl'IIIl'J'l-'- Hobllu clil It, 1881 Vulatlna ----- F,.uQU~ot 01)1 .. 1183V'uh::.nD (Wetl -_ F,ID.lttlll II., 100;' Rhvolll... - Wohll!ltl,18631Ofuiln.al fPI1•••1optlnllllnl- CSe" at1ld 6p.trIU" 1918 8111..111 WohliMZ., 1,9831tl,a1ruJ tphrulDplhtJ..n) ••••_ S.lt .nd Sp.rk., 1978

..·3

FI G U R E 2 (a) Grain size distributions for phreatomagmatictephra. (b) Clasts shapes characteristic of phreatomagmatic frag­mentation.

Theoretical 3nd experimental studies at :India Nati naJLaboratory ew Mexico) in ci1 I 70 and 1980 pl' ­vided a trong basis for understanding ;CIs and a &3J11 ­work fOT the smd of phreatOmagmatic activi . Th e

PHREATO.\IAG.\IATIC FIlAGMEI'.;TATIO:-l 435

experiments were designed to investigate the occurrenceof stcam explosions during core meltdown at nuclearrcactors and others industrial environments such asmctal smcltcrs. The primary goal of the experimentswas to const:r:Lin models for the conditions for a steamexplosion when a coolant comes into contact with a fuel.Fuel (melt) used in the experiments ranged from moltensalts to themljce-gener:ued melts. Results from thesestudies include the four steps describing the physicaldynamics of a FCI.

Phreatomagmatic explosions are complex interactionsof four phases which are often i.ntern:'!lIy heterogeneous(magma, magmatic volatiles, extern:'!1 water, and wallrock). Experimental studies ofsimple systems offer pow­erful tools for understanding the eruption dynamics.

B. Thcnnitc: Water Experiments

Experiments conducted at Los Alamos National Labo­ratory were designed to simulate phrearomagmatic ac­tivity using an Fe-AI thermitic melt and water. Theobjectives were to monitor dynamic events and to deter­mine what controls me rapid conversion of the melt'sthemlal energy imo mechanical energy when it interacts\\lith water, The melt, simulating b:lsaltic magma, re­sulted from an exothcnnic reaction of fine-grained alu­minum (-24 wt%) with magnetite (-76 wt%).ln gen­eml, this thermite composition yields 3bout 1130 kJexcess heat per mole of iron oxide at J800 K The physi­cal and thermal properties of Fe-AI thennite melt arecompared wlm those of a typical tholeiitic basaltic meltin Table L

Four designs, sha"..n and described in Figs, 3a-3d,were employed to evaluate the effects ofcont:lct geome­try, the mass ratio of water to melt, and the confiningstrength (or pressure) on the explosivity of the experi-

TABLE r Physical and Therma.l Properties of Fe-AIThermite and Basaltic Melt

Proptnies F~AJ melt Basoltlc melt

Uquldus T 1000-2000 K 1370-1520 K

Enthalpy 3700 kj/kg 1150kjfkg

Viscosity l~ Pa s 1~-101 Pa s

Density 3.0-1.0 Hg/m l 2.5-2.7 Hg/m l

Surface tension O,S N/m O,)S N/m

Thermal conductivity VI J/(m s K) 2,1 Jf(m s K)

mental phrearomagmatic eruption. High speed cameraswere used in all experiments and samples of ejectedthennite were collected from some of the experiments,

More than one style of melt ejection, each resemblinga type of n3tural volcanic 3ctivity, resulted from these ex­periments:

I. Continuous fountaining ofcentimeter-sized liquidfragments of them1itic melt, which resembled a hawai­ian eruption

2, Pulsating ballistic ejections of partially quenchedand vesicular (scoriaceous) fragments, which resembleda strombolian eruption

3. Dry v:tpor-rich explosions ejecting micrometer­si7.ed fragments in expanding jets of superheated ste:.J.m,which resembled a surtsey:m eruption

4. \.vet vapor explosions ejecting miUimeter·sizedfragments in ballistic plumes of condensing steam

5. Passive chilling of the mermitic melt il1to centime­ter-sized fragments with quenched surfaces, which wasanalogous to the formation of pillow lav3S or peperites

A fifth experiment was designed to resemble a propul­sion rocket (Fig_ 3e) in order to quantify the W:'lter-to­melt mass mtio and confining pressure controls on themechaniClI energyofan imeraetion.ln this design, ejec­tion of high pressure steam :lI1d fragmented themlitepassing through the vent caused the vessel to lift off theground. Measurements of the internal pressure historyand ejecta chamcteristics were recorded, as was the maxi­mum height reached by the vessel. Three confiningpressures were used, 6.8, 16.3, and 35.7 MPa; the lastexceeds the critiClJ pressure of water (22.0 MP:'!). Thew:lter-to-melt mass ratio investigated ranged from 0.38(1.82 kg of watcr and 4.73 kg of melt) to 2.00 (4.00 kgof water :md 2,00 kg of melt).

The results from these experiments include three dis­tinct pressurization histories or groups which werefound to correspond to three specific ranges of lift­off heights (LOH) and produced clasts with distinctmorphologies. The three pressurization histories arecharacteri7..ed by (I) exponential pressure-time curves;(2) an initial Hnear pressure rise and aftcr 3pproximatcly0.3 to 0.8 s, a parabolic pressure increase to :1 v:llue nC:lrthe confining pressure; and (3) an initial linear pressureincrease followed by an event th3t pressurized the cham­ber to a maximum v:tlue within 5 to 20 s.

Although the number of experimental samples is lim­ited, an intercstingcorrclation was observed between thesix distinct grain sh:'!pcs (Fig. 4-J) identified in reco\'eredsamples and the three pressure groups: Group 1 experi­lllents produced predominantly l1losslike grains; Group2 produced mainly spheric:ll or droplike :ll1d ;Iggregate

436 PlIREATOM GMATI FRA I' ATJO

grain . Group 3 e 'Perimen produced main] blockand irregul:lrly shaped grains. Platelike grain werefound in tbe fine-g'J'ainfraction f mos sample. Thecorrelation betwe n grain shapes and tbe dlree pre suregroup sugg tha the grain are a cons quence of thedynamic behavior of water a it i being heated b themelt during prebur interaction. though sarnplingdidnot adequately repr sent the fraction of the sample -finerthan 62 nun (4 cp), the di tribution hown in <ig. 4bare ry similar to orne phreatomagmatic tephra izedistributions .ig. 2a).

In rder Facilitate tb e of these e 'perim nts as3mllogs of hydrovolcanic erllptions of much larger scale,

the explo we energy is expre ed me ra of themeasured and calculated mechanical energy co the initialimernal (thermal) energy f dle melt. This rati is calledtb conver ion ratio (MeR) and wa calculated fr mthe conservation of momentum.

shown in Fig. 5 a po ·ti e correlation was foundbetween CR and the water-to-melt mass ratio (Rm)

within the three pressure groups. Thi plot h ws thatthe obs rved pres Ul'jzation III de is tJ'Ongly Hnked t

th resulting con ersion ratio for each xperiment. Thpressure- .me hi ton associated with each pressurgroup reflects the degree to which the w3ter-thermimelt interaction approached thennal equilibrium. Thus,

(a)REACTION

t1m

Iron Cas;"g

Sane!Thermlte

(d)Vent Opening(9.8-<:m diam.)

3-an

VenlPipe60cm

AJumino"mPar1iIlon

36-an i.d.

.........,~ii~l!ifi1 ~~'l-- SUl'llt Valve~1000 psi(Ej6illm)

Press.uro andTempernllJre

Gaugl!5

Thermite

_AluminumPartition

"'-H 0z

WElBr

Eu~....<0

PressureI=~""'~ Transdue::er

Sand

I~~,.+ i:'--lgnilicn Wira

(e)

r-;;:~~:::>-lgnitian\IiThenn1teSand and H20

MagnesiumParlition

~O

,-__ -Ignl~onW1ro

R----F=~t-::::::===R- ~al BuckslPIeX\;llass

~o

(b)

(e)

'\Bursl Valve

FIGURE 3 Schematic drawings of the five thermite-water interaction experlmenLS conducted at Los Alamos National Laboratory.(a) Emplacement of 10 kg of thermite in an iron pipe above a second pipe tilled with water: (b) immersion of a Plexiglas tube in a largePleXiglas box filled with> I cubic meter of water; (c) emplacement of 90 kg of thermite held above water by an aluminum disk In asealed neel cylinder with a vent at the top that burst open when the Internal pressure exceeded 7 MPa; and (d) same as design (c)except the vent pipe extended down through the thermite into the water compartment. (e) The lift-off experiment designed similarlyto (d) except the burst valve is located at the base of the vessel.

PHREATOJ\lIAGNIA'll F loi -NT TT N 437

A B

(b)

(c)

(a)

9 10

7 8 9 0

SAMPLE • LOF- 1

SAMPLE .lOF-l

SAMPLE. lOF-13

~':::<.. .. ..l'I'~'i'I'iiii.

·5 -4 ·3 ·2 ·1 0 2 3 4 5 6 7 8 9 10PHI SIZE

·5

2Sa

a

P1ate-Uke

Blocky,Equant

oss-Llke.Convolute

Sp $. EDrop-l.lke SF

~!!I.iii~ ~0.L........,rr-r.....r;~i=TT'i-HL,:>r,r--r-~.......-,r-T""T'"T""T""T'"T

FI GU RE 4 (A) Sketches of the four grain shapes observed in samples collected from the LANL experiments: (I) blocky; (2) mosslikeconvoluted surfaces; (3) spherical and droplike; and (4) curvey. plate-like. (8) Results of sieve analysis of three samples of experimentaldebris. Sample characterization where ep = - log2 (diameter in mm) includes: (a) LOF-\ with mean 11 = 1.74 and standard deviation(0) = 1.45; (b) LOF-II with mean ep = 1.25 and (J = 1.99; and (c) LOF-13 with mean cP = 2.19 and (J = 1.90.

. of thermal explo ion experiments (TEE) at ni­ve i f uerzbw-g \ ere designed to rudy phreato­magmatic fragmentation using remelted volcanic rockampl with compo itions ranging from nlrrabasic to

water-Ill It interactions iJ1Volved with Group 1. experi­ments were th m st efficient and explosive, aJ1d camecl ses t thermal equilibrium. These equate to "illy"phreat ma mati ev nts. Water-melt interactions in­volved with r up 3 xperiment- were the least efficienand explo ive and did not approach thermal equilibrium.The e exp rimen were con idered analogs to subma­rine phr at magmatic events. Group 2 experimen fellbet\ cen r up 1 and 3 and were considered analogst phr~tomagmatic e en .

andesitic. The experiments evaluated the effect of themod f ntac between water and melt, the mixingonditi n , the fragmentati n processes during 3J1 ex-

plo ive ) which i :lis termed a thermohydr:lUliexpl i 11, and the expan ion phase of an explosive FCl.

1experim n involved high resolution measurementsf phy i aJ pr e e and properties a ociated with an

ex'})l ive F and th e aluation of fragmented m It.The e experiments included two con1igurati n ':

1) water entrapm nt achie ed by injection of a watervolume int a mel olurne and (2) melt entrapmentachi cd b th tification of water and melt on anxperimental Ie of 5- 0 em (Fig. 6). The latter pro-

duced nl ery weak I with low reproducibili andthus ~ er n eful analogs to phreatomagmatic explo­ion. h former wer useful as these produced ener­

getic xpl i n ry ffecti el . For timing purpos anexplo i n in each e.'-'J>Crimcnt\l a initiated b a projectile

tilizing Rock Samplesc.

438 PHREATO.... GMA I FRA ME. ATI

o

2.0

P-III

Re ul fr m me first twO experimental designs ug­gested that heat I . from me melt during me initialcontact an pr mixing pha e i critical for the generationof an e''P1o i n. Two c nract mod were in estigatedhot mel surr und d b exc water or water sur­rounded b exc melt. vVhen the melt was disp rsedin exc \ ater the h at tran fer e en under stablevap r film boiling conditions wa Vi ry effective suchthat m lt partiel \Vim diam ters <1 em completelolidified or a mick crust formed around larger melt

partiel wimin <1 . In tb.i case explo i e FCI did notcur. In the e of \ ater encrapment by exc melt,

an e;\-"plo .ve' T, f, und to be highly probable. Herethe in en i of a r ting thennoh draulic explo ionis proportional to m initial contact areas.

The mixing phase \ found to be the primaryconcrolon rh in ensi of an :plo ive CI. The available in­terfucial area b t\ een the two liquids is governed b(a) the differential flo sp ed (hydrodynamic energy)bet\ een me 0 fluid and (b) b th material propertiof the fluid. fficient mixing (i.e. creation of a largeint rfacial area in a hort time) " as achieved hen thflow peed i high and th densities and visco'ri ofthe liquid ar imilar and 10 • ilicate magm havedensiti ben en 25 0 and 3000 kg m-3 and vi co iriranging fr III few Pa to >106 Pa . In contrast, waterha a den ity near 1 00 kg m-l and a vi co ity of 10-3

Pa . In .jg. 7 and the optimum mixing cOlldi 'oosfor I w vi i l ultraba ie melts are the maximum val­u for th W3.ter-to-m I volume rario and the differen­tial fl w sp cd which c ITC pond to 11 vo]% water and

• 6.BMPa

°16.3MPa+35.7MPa

o

8

8.----.----,----.----,-,...-...,.....--,----.--.----.---.---.-----,7

6

dlat wa fired a the m It surfu to breakdown insulat-ing ap r filn .

third >., erimen v d igned to identify and quan­tify fragmen p pulati n generated by proees e a oci­ated widl dle xpan i n uperheated steam ( rage 4of:1 F 1). h .:perilUen in olved a udden releaseof 3 c mpre in r (argon) in dle melt filled cruci-ble (Fig. 6).

Am

FI G U R E 5 T rend of mechanical conversion ratio (MCR) withthe waterlmelt mass ratio ( ) for experiments with differentp..,...The delineated field shows the range of data for the experimentsexhibiting pressure records of a given group. P-I experimentsshowed MCR values greater chan those of P-II. and P-III experi­ments showed the lowest MCR values. This relationship stronglyindicateS that the mode of pressurization is linked to the mecha­nism of fragmentation and heat exchange for chese experiments.

~ 100e....~rJlI::Q)

75.5I::0'iii00. 50)(CD

"Q)

.!:!ai 25E0z

00 10 15 20

Water/melt volume ratio

FIGURE 7 Waterlmelt volume ratios of mixing in entrap­ment configuration. The melt used in chis experiments was aremelted olivine-melilitite at 135 C and a viscosity of about IPa s.

DcBAFIGURE 6 Entrapment of water in melt. Schematic crosssections of crucible used in the experiments: (A) A thin walledsteel tube is Inserted Into the melt to the desired depth. (B) Wateris Injected into the melt; thus a local waterlmelt premix is pro­duced. (C) After triggering. the escalating process of thennohy­draulic explosion takes place in this premixed zone. dUring whichwater is superheated. Once the superheat reaches a critical valuethe phase transition to superheated steam is enforced. resultingin very rapid volume expansion. (D) The superheated steamfinally expands to ambient pressure in an adiabatic way. chusaccelerating the overlying melt strata out of che crucible.

PHREATO)l;1AGMATIC FRAGMENTATION 439

100~

"e..2.-.z.'iii 80 -CQl

:sc0'iii 600

~Ql

"0Ql

.t:! 40OJE0z

exclu. ively produced during this event were identi­fied and quantified. The total of new sill'face area wasfowld to correlate linearly with the respective explo­sion intensity.

III. FRAGM ENTATION M ECHAN ISMS

A. Types of Fragmentation

FIG UR E 9 The collapse of a superheated vapor film or theexplosive expansion of the film will produce stress waves in themelt. If these exceed the bulk modulus of the melt and It fracturesin a brittle fashion, (a) blocky Type I or (b) Type 2 pyroclastswill form. [Reprinted with permission from Wohlen, 1983.]

A fragmentation mechanism associated with each phrea­tomagmatic grain type (i.e., blocky, fusiform, moss-like,platey, and spherical or drop-like) can be inferred&'omresults of the phreatomagmatic analog experiments andFcr theory. The blocky grains found ill tephra depositsand produced in lab experiments resemble hat1:eredglass or glass that has been partially twisted and de­formed willie still plastic. These &'agments form whenstress waves propagate through a melt and produce de­formation rates that exceed its bulk modulus (Fig'. 9).Also it is envisioned that the local temperature of themelt is near solidus and brittle fractures develop in re­sponse to tllermal contraction or tensional stress wavesproduced by vapor film collapse. Because stress wavestypically propagate faster thall tllemlal waves, it is likelythat tllermal contraction affects only melt smfaces wIllie

STRESS WAVE FRAGMENTATION

Brittle FractureVesicu lalion

--r-----r--

___1 ~_~ 1 6P6T

~,~~~~~~~~ed11'111///1 '111//1/'11/11 ........ {T'" 370"<:

Melt P '" 20 MP.

_..L ..L __

-T-----,--

l:o-5oo~m

Quenching

~~tw~~~r¥--vapor Expansion

20 L-.L---'----'----'----'----'-'-'--'-..........---'---.J.......J..---'----'----'"-'--'-'-'--'--'---"-----'

o 2.5 5.0 7.5 10.0 12.5Differential flow speed (m/s)

FI G UR E 8 Differential flow speeds of mixing at a water/meltvolume ratio of I I% in entrapment configuration. The melt alsowas remelted olivine-melilitite at I350°C and a viscosity of aboutI Pa s.

-4.5 mis, respectively. These values are thought to bevalid for all basic magmas for cases in which water in­trudes on the magma. Although these maxima are welldefined, explosions occurred in a wide hydrodynamicmixing range.

Three fragmentation regimes were ilistinguished:

1. Fragmentation ofthe melt inside the crucible with­out measurable expansion of the mixture during 6Jle­fragmentation and heat transfer (Stage 3 of a FeI)

2. Fragmentation of melt strata dw'jng accelerationof the melt by expansion of superheated steam insidethe crucible

3. Fragmentation of the melt during high speed bal­listic trajectory outside the crucible

Particles produced by the fine-fragmentation, regime 1,resembled the ash-sized particles generated by theL.ANL experiments. The fine-fragmentation phase wasfOUJld to be the cmcial phase for an explosive FCr,during which the maximum kinetic energy (explosionenergy) is released. This phase produced angular shapedparticles with diameters between 30 and 130 .am. Frag­mentation during melt acceleration in a confined ge­ometry produced spherical fragments with diametersbetween 90 and 180 .am. Fragmentation during deceler­ated movement of melt on ballistic trajectories in freeair produced elongated shapes to hairlike shapes similarto Pele's hair, depending on the ejection velocity.

An experimental explosive FCr is a highly energetic,short-time process occurring on a timescale of < 1 ms.Within this time [Tame, new surfaces were createdby a brittle type of fine [Tagmentation. The partides

440 PHREATOMAGMATI FRA ME rTATIO

the bulk of the melt is rapidl ubjected to mechani­cal 'tr

In qu nched portions of th mel fractures tend topr pagate at mules I than 45° 0 the melt surfacefI rm.in bl ck-shaped cIa ts with pI nar to curviplanarwfac. p r film oscillation ma pr duce turbulence

that ten to spall quenched mel fragments from thesurfac of th molten body and exp unqu nched sur­faces (Fig. 9). For portion of the melt that fragmentpri r quendling 3por turbulence can cause defor­mation f unquenched shards \~~th mel ur['3ce tensionprom ring forlllation of fluidal surf.1ce e"-'1:I.lres.

M s-like grains develop by vi. c u deformation un­der ten i nal sU'css conditions created dUl"ing rapid va­por 6 I'mation. These conditions are predicted by thedev lopment of Rayleigh-Taylor and Eelvin-Helm­boltz ins abilities. These instabili 'es are caused by therelative motion of the vapor along the melt surface (Fig.10). h perturb the mel unsce and introduce addi­ti n31 morphological characteri ti . Tin plumes ofmelt ri from the snrface wh n rface tenSion forcesare xceeded b Rayleigh-Ta lor in mbiliti (Fig. 11).

elvin-Helmholtz lnstabiliti ar induced by heartr e dlat furthertrerch :md di p the melt sur-

fuce. B cause th abilities fonn with a range of

wa elengths and oriematio melt nrfaces can becomery con oluted resulting in t rtuous high urfac

31" fragments.pherica1 or droplike graill d elop from the effect

f surface tension when magma i till hot enough tohave like a fluid. with rno - haped grains rn It

surf3ce instabilities both grO\ and d tach to fonnspher­ical or elongated ribbon-lik structUres in response tothe force of locally collapsing V'3por films (Fig. 12). Them t rapid mechani m contri uting to production ofdl e grains oCCW'S when compres ion waves associatedwith vapor film collapse becom large enough to hatterthe surface undulations produced by instabilitie ; thiresults in the production of a multitude of fine-grainedfragments. When axisymmetri collapse of vapor filmsproduces accelerations sufficient to overcome sur[-acetension forces offluid melt, wat r jets can penetrate intothe melt and become partially or c mpletely entrappedbeneadl the urface. The trapped \ ater b comes super­heat d and expands thus causing additional fragmenta­tion of melt.

Platey grains have been in erpr ted as pieces ofquenched crust stripped. ff the surfuce of dle magma

ig. 9b). he stripping proc can b attributed 0 theturbulence of vapor film oscillation aporization wa e

FLUID INSTABILITY(PLANAR TAYLOR MODEL)

Collapsing Vapor Film tfe.- A1110 K-----I

I-- ), crll----l

:,:.,).:: ::.Mean Fragment

Diameter

Waer---t---

Vapor"... ., ..• ~......

Magma

. .. (Non-Growing

---t----

~

T'lmOK

1

-0 -:-"0 "-CO--::J Vapor ando 0°. 0-()O .. 00 0 00 Fragments

~

One Cycle of Surface Wave Growth and Detachment

FIGURE 10 (Top) Sketch of a planar Taylor instability at the interlace between magma and a collapsing vapor film. (BotLom)Diagram of a complete cycle of Instability growth. The oscillation of the vapor film transmits sufficient momentum to the magmasurlace so that it becomes distorted into waves. [Reprinted with permission from Wohletz, 1986.]

PHlt£ '1"OM,\ MATle < RA AruN AT10N 441

pr p:lgatio or Kelvin-Helmholtz. in tabili . High e­loci ejecti n of liquid mel leads me fonnation ofrock wool.

lind hence ill beat ttansfer. The intenningling is clri enby diffu jona! and kinetic mechanisms, includin~ h. ­draulic fracturing tress corro ion capiUarity apor­film coUap e and water jetting, m lecuJar diffusion and

aytor and Helmh Itz instabiLity.l is imp rtant t notethat dlis intermingling occur over a wide range ofscales6- III submicromet rs to meters, and that i ceu with­out si!!1lificant hC3t tran er to the a r because ofapor-film insulation.

After premixing, thenn31 energy is tta .ferred t thwater over a large effective surfa e area, and with thistransfer, pressure builds in many ClSes m:1king the watera supercritical fluid or a metaStable superheat d Ruid_'rom :perimen this pha'e of pr growth maohow se era! different t mporal panems including

ponentiaJ, parab lic and cyclic gr wm , ith tim . Forexponential growth pressure increases with urface areadwing premixing, which requires that prem.ixing pro­due mel fragments that d ere e in i7..e linearl wimtime. Thi produc an exponential ri e in rface area,hence in hC3t tran fer with time and an accompanying

onential ris in water pressw·e. On th ther hand,a]larab Ii pres ure rise probably r fl cts imple thermaldiffusion in which heat transfer i initially rapid and10\ with time a diffusional di ranees iner ase. yelic

pr sure increases b reR ct cyclic gro, and coil pseof apor films a pr mixin progr . Rapid hC3t rrans­f; r at a water-melt interface cau es a vapor film topressurize and rapidly exp~tnd past its sta hie size, ladingto overcooling of tbe film and its immediate condensa­ti n and coUap e, ' hich ma pr duce " 3ter je that

Mossy Grains

Vapor Explos on

Co-penetra 'on of Melland Water

S pemealedVapor Film

T"" 700·C{p", 35 MPa

Film Collapse_ _ .Fluid Instablity-Water

~~~~~~~I~I: '. Jet MixinglJ/ High Surface AreaII Particles

FLUID INSTABILITY FRAGMENTATION

2.

1.

FIGURE I I Auld instabilities can form at the contact ofwater with melt. These insubilities are of a Rayleigh-Taylor orKelvin-Helmholtz type, both of which may cause axis-ymmeoicfilm collapse that causes water jets to penetrate the melt. Theresult is high surface-area fragmentation, rapid heat exchange,followed by vapor explosion. The explosion then generates newcontact areas and/or a shock wave that fragments more meltand acts as a demnation wave. This process requires abundantwater m be present. [Reprinted with permission from Woh­lea, 1983.]

FLUID INSTABILITY FRAGMENTATION

B. rowth and Evolution ofFragment3 .on E eots

1.

FIGU REI 2 Fluid instabilities will fonn spherical or drop­like shapes of melt if viscosity is low and surface tension effectsare strong. [Reprinted with permission from Wahlen, 1983.]

Phreatomagmacic fragmentlll::ion e olves in a number ofdifferent manners d pending upon th primary control­ling parameters of viscosity, temperature, confiningpre ure and water/magma contact mode (i.e. mixingconditio and mas ratio). Th e parameters are larg Idetermined b the environmental e .ng of m oleanand itS magma charaeteristi ,discuss d belm and dll~}'

det rJlline the rate of heat transfer from tbe melt towater. Fragmentation events follow a complex growthhistory that includ premixing, pr ure growrh ttig­gering failure and ej ction, hese are imilar 0

tho a oeia d with an FCl.Premixing is an increa ed int rmingling ofwater widl

the mel, which lead t a larg'e increas in surface area.,

2.

Drop-like and spherical grains

Vapor Explosion

442 PHREATO,\I,\GMATJC fRAGMENTATION

-

impinge the surface of the melt. The initial film pressurethen becomes the force exerted on the melt surf.lce bythe film's colhlpse, which leads to the eventU<l1 breakupof the melt into smaller particles and :lt1 increase in heattransfer surface area. As the surface area grows, thepressure of each cycle of film growth also increases.

Premixing and pressure growth may continue or maydecline with time, the latter case resulting in a highlymetast:able situation that can exist onJy in high confiningpressures (> 16 .M..Pa). However, for the former case,water will gradually increase in temperature until itspontaneously vaporizes in wbole or part. If a smallportion of the water reaches this point first, its vaporiza­tion produces a pressure pulse that triggers the vaporiza­tion of the rest of the water and may cause an explosion.

An explosive vapor expansion drives a shock wave bywhich the melt is rapidly acceler-lted. This accelerationproduces stresses that can exceed the bulk modulus ofthe melt, leading to its whole-scale failure and produc­tion of fragments; fragments formed during premixing,lre further comminuted and intact melt is torn apart.The mode offailure is largely detennined by the proper­ties of the melt, and it ranges from brittle to ductilein behavior.

The final stage of fragmentation is the ejection phasewhere fragments are moving at high speeds. If the explo­sion originates away from a free surface, then as theshock wave emerges through the free surface it wilJcreate a rarefaction wave that propagates in the oppositedirection. This rarefaction wave allows immediate vaporexpansion as it propagates into dle conduit. It is at tltisstage when the wall rock (nonjuvenile lithics) becomesfragmented and incorporated into the erupting mass.With vapor expansion, fragments are accelerated andejected out of the system. Duringthis ejection, the speedof the fragment is proportional to its size because ofinertial effects, such that fragments will move at differentspeeds and collide with one another, leading to furtherfragmentation.

Melt fragment shapes and size distributions are con­trolled by the types of premixing, triggering, failuremodes, and ejection. Although growth and evolutioncannot be direcdy observed, the overall pattern of phrea­tomagmatic fragmentation is portrayed by eruption phe­nomena. In this light, fragmentation evolution can beclassified as: (I) escalating, (2) declining, (3) cyclic, or(4) delayed. Escalating fTagmentation during eruptionresults in larger and larger volumes of tephra producedin each successive burst. Such behavior requires the massof magma fragmeilted to increase by an increase in thewater flux and/or mab'1l1a flux. As discussed ,lbove, thewater/melt mass ratio (R,,J detennines in part the energy

of fragmentation with higher fragmentation energyleading to more magma fragmentation. Conversely, de­creasing fluxes or a change in Rm away from high energyfragmentation leads to a declining fragmentation evolu­tion. Cyclic evolutions are common where fragmenta­tion repeatedly grows and then declines. This situlltionis most common where water supply is depleted afterlarge fragmentation events, resulting in less energeticones until the water supply is reestablished. Delayedfragmentation refers to a situation where II fragmenta­tion event occurs after a prolonged period :ltter pre­mixing. Prior to fragmentation, the system is metastable,requiring a trigger. This type offragmcntation, althoughwell-observed in experiments, is difficult to sustain involcanic situations. Such situations might include a up­per-crust magma chamber that interllcts with water saN­rated rocks for a long time prior to eruption.

C. Fragmentation Controls

Through the experiments llnd theory described aboveand observlltions llt numerous volcanoes around theworld, it hlls become clear that phreatomagmatic frag­mcntltion is largely controlled by three key parameters:magma viscosity, temperature/pressure, and water/magma contact mode. The dlird is a function of thesupply rate of magma and external water which is deter­ntined by the hydrology of and around the vent :mdconduit.

During phreatomagmatic fragmentation pllrt of thethermal energy of the magma is converted to mechllnic,llenergy, including seismic and llcouStiC energy, fragmen­tation energy, and kinetic energy of fragmcnt motion.With an increasing efficiency or conversion rlltio, moremechanical energy is released and fragmentation in­creases. In field observations at volcanoes, the relativeefficiency of phrelltol1lllgmatic fragmentation is re­flected by £rugment size distribution and mode of frag­ment dispersal.

In general, (I) higher temperature melts have greaterthermal energy available for conversion to mechanicalenergy, (2) viscosity retards the mixing of magma andwater in a manner that generally predicts th:lt fragmen­tation will be greater in lower viscosity magmas, and(3) the amount of water relative to tllat of mllgtlla in­volved in fTagtllentation determines the intermediateand final thermodynamic states of water for phreato­magmatic eruptions. Magtna and water tempernturesprior to interaction influence the heat transfer rates andequilibrium temperature a.pproached during mixing. Ina simple way, the higher the equilibrium temperature,

PHREI\TOMAG.\IATIC FRAG...1ENTATION 443

the more energy is avrlilable for mechanical work, butthe r;)[e of heat tr:lnsfer is complexly linked to watertempef'3ture, rising r:l.pidly with tempef'3ture until theonset of vapor-film boiling. Thennal equilibrium isprobably never reached during fragmentation becauseinsmbility and vapor expansion occurs before this pointis reached. For the instlntancous heating case, the tem­per:uure of the interface, To, is theoretically con­str:lined by

T-<a.r';;.) + T.(a,./V..)T. ~ (a./V..) + (a,./v..) ,

where 0' and K arc thennal conductivity and diffusivity,respectively, ofthe respective phases water, w, and melt,111. If T,. is greater th;lll the spontaneous nucleation tem­perature of water, then a spontaneous superheat vapor­iZ3tion would occur.

The effect of :unbicnt (or confining) pressure is lessunderstood. Pressure is thennodyn3ntically Ijnked totcmpef'3ture, and, as such, the pressure-volwne work ofa fragment:ltion should increase with pressure, but thereare some experimental indications that pressure also actsto subdue interaction be<:ause it limits vapor-fiJm growthand theamountofpremixing. However, the most,riolente);"pe.riment:ll fragmentations have occurred where con­fining pressure is in excess of the critical pressure ofwater.

The viscosity of magmas ranges from few Pa s in thecase of basaltic composition to milljons of Pa s in thecase of silicic magmas. Viscosity is addjtionally coupledto tempef'3ture in m:1gmas (higher temperature magm:1Sh:1ve generally lower viscosities). \¥hiJe temperaturemay vary several hundreds of degrees, viscosity variesover many orders of lI1:1gnitude. Viscosity (and surfacetension) is an import-Jilt p:lr:l1lleter tor growth :lnd prop­:lg:Jtion of fluid instabilities :It the interf:lce betweenlll:lglll:l and W:lter. Such inst:lbilities have been cited :IS

likely mechanisms.The combined effect of viscosity and temperature on

the efficiency of fragment::ltion is markedly less th3ndl:1tofw:lter/magma ratio, which varies over lWO ordersof magnitude. This observation reflects the major con­trol of w:lter expansion during fragment::ltion, a controldetemuned by intennediate and final thennodynamicscates of water. If very little water is available, it can beheated to high tempcraNres and pressures, but frag­ments onJy a relatively small volume of magma. At theother extreme, abund:lnt water may never get enoughthennal energy for fragmcnt::ltion. At intermedjatc massratios, there is enough menual energy to take all dleWolter to high states of intern:11 energy, thus resultingin rather complete magma fragmentation.

Water involved in the inter.tction may not always bea pure liquid, it may cont::lin particles (sediment ortephra) or other impurities such as dissolved compounds(seawater or hydrothennal fluids). The effects of theseimpurities on dIe fragment:ltion process are related tothe changes in the physiC31 prol>crties of the impureWolter. Depending on the concentration of impurities inthe water, its viscosity may increase up to :111 order ofmagnitude, the density may double and the heat C3pacitymay decrease by one-fourdl.

IV. FIELD OBSERVATIONS

Hydrologic settings tor phrcatomagmatic t"r:Jgmcnt:ltionvary from deep ocean (marine) to ncar shore (littoral)to lake (lacustrine) to surface drainages (fluvial) andgroundwater (phreatic) settings. The hydrological set­tings affect the abundance and supply rate of water andthe ambient pressure 3t which interaction occurs. Ac­cordingly, the cha.racter ofphreatomagnlatic fragment:l­tion varies in a systematic manner with setting.

For marine settings, deep water explosive fr.tgment:l­tions are not generally observable. In bet, many re­searchers propose that for depths where hydrost:lucpressure is greater than the critical pressure of water,explosive fragment:ltion does not occur. These research­ers explain deep ocean hydroclasts to have resulted fromsimple quench fragment:ltion of extruded lava, becausemost hydroclasts are blocky sh:lped and coarse to me­dium in size. On dlC adler h:lnd, experimental and dIeo­retic:ll evidence shows th:lt the critical pressure of W:lteris not the m:lximum pressure at which explosive fr:lg­mentation m:ly occur. In t~lCt, for most subStances th:ltundergo such rapid vaporization (leading to explosiveexpansion), the phase change may occur well above thecritical pressure. Future research involvingacru:ll obscr­v:Jtion of deep water eruption or documentation of ex­plosive effects in deep water tephra deposits may resolvethis debate.

Littoral fragment::ltions have been well documcntedin Hawaii where lava flows enter the sea. A large '':lri:J­tion in fragmentation dynamics is observed from passivesteam and quench-fragrnent:ltion to violent bursts thatform high-speed Spr:J)'S of fine-grained hydroclasrs. Onecontrol is thought to be whether or not laV'3 contactswater \\~dun dle confined space of a lava Nbe or not.The fragment:ltioll st:lrting in a lava rube is observedto be very ,~olent, suggesting dlat such confinementallowed buildup ofhigh pressure during the event. Con-

444 PHil· TO .'1 TIC F 1£ :TATI ~

equently, fragmen from lin roll fragme~tation ~) ,a wid range of sh3pe and size. from c ntJmeter- lZ

Iluid- hap d clasts to microme er-sized blod.'}' shaJ:d .Cone' that form in the lin ral zone are imilar to tuffcon that grov in lacustrine environmen .These erup­tions ar ob erv d produc large am un of turated

earn and are caU • \ et eruptions. h associa cdcon :11' referred a littoral one or r d cratefor tll that f, rm where laY'd bas tra eJed 0 er a mal' hor a b g.

PhreatomagU1atic fragmentation in fiuvi:ll or plli'eatihydr g logical environmen i governed b a limj edsuppl f \ ater. uch fragmentations thou~h, can ~e

th III violent bCCluse , ater/magma rano are 10

th ran c of optimal interaction and the contact munder nwed nditio (i.., in hydrauliClUy ac . efault t m ) ar ptimal £. r the generari n of dlenn ­hydraulic explosi n. Tuff ring' (e.g., . ratcr Elega.llte,Mexi OJ PaJlUm crater, United States) In these settmghow wide-area hi h energy dispersal with uperhea ed

vap r that do n t produce such wet d p. i~' hen epaJa nirization i limited. uch ~gme~tanon 1. calleddry and in man plac occu \VIth cona depoSl thatresuI fr m little l' no water interaction. ypicalJ thefragm n are fine-grained a h, showing a. range fshap s fr m bl ky to mos aggregates t n1Jcrom ~ 1'-

ized 'phere .Tuffc nes (.g., erro I rado, MeXl ;Koko rater Uni cd tate) sh w low energy dispel' alwith :lturated ~ am that produce palagonirired dep ­its. u h fragmentation is called et and in man. pI:!produce basaJ surg deposi that result rna progr -sivel \ ater-rich interacti n. TypiCllly the £ragmenare 6ne- to medium-grained a h, howing a range ofshap fTom bl k')' and platey with du t- ized fr3g­ments adhering' t surfaces, t mung ac l' tionmy lypegrains. .

The depdl of fmgmentati n is likel termmedthe depth of the aquifer. Durin eruption an aquiferrna be drawn d wn such that fragmentati n level gedeep r wbile the hydrostatic 1'1' e n~ the f, mis m re r less c n tant. Expl ions pI' pa ring down­ward will fragm nand erup wall r ck at eeper anddeeper levels, dlU' enlarging the conduit (e.g., a .di~­

treme) 0 the depth f the aquifer. The am un of IJdllejecta also reflec he ~'t:abili of tile wall .. ery Ii efragmen ed aU rock i erupted when the aLi are tableand the erupti n' fairl continuous and \ eak. If thwalls are unsrabl then the \ al rna be roded durin...a \ eak eruption thus producing a ignifiClnt amountof fragmented v aJl rock. The coJJapse f eroded ,vallrockc:1Jl trigger a tr ng, di I' te explosive phreatomag­marie 1'uption thac erupts m stly lithic material.

V. SUMMARY AND

FUTURE RESEARCH

1. Phreatom:u!Tnatic frJgmentation i quite d" tinctfr m magmatic fragmenta . n in both me erupti\·c pbe­n mena and th driving m chanisms well as in metype of fragm n produced. Phreat magrnati m pro­du e unique juv nile grain populati n that rdle t theI'elati e 3l110UJlt of water and magma involved in theinteraction. he quantity f nonjuvenile (fragnl med\ all rock) material found in depo i reflect the iliryof the condui ,all and the explo ivi of the rup ·on.

. Phreatomagmatic frngmenration l' ults in:l pec­trum 0 erupti activity ranging from p:l i e quenchin lT

f magma/la a t eA'Plo ive ejectio f:l h. The tyle oferupri n primarily refiec . th hydr I gic envir \lmentin which magma/lava come into contact with water. .

3. Much of ur wlder tanding f phreatomagmaocfragm Dtation tems from laborato .. experimen . re­lated to fuel-co lant im ractions a )a weU:I fieldob enations and laborat anal' f the £ragmen .I . furor eA-penmentaJ rodi that ma}' b t improveour current kn wledge or dle contr f phreatomag­mati m. For example, xperimen ha e thus far nlyc vered a rarllcr narrow range of mel comp i' n ,inluding thermitic melts and om r melted v Jeanicrock. etailed experimen . may reveal how fr-Jgmenta­tion i affected b comp itions ranging from mafic

ilicic and with different proporti ns of di I edv law . imilarl experim oral coolan have beenpure or near! pure \ ater. 0 very little i known abo uth \ impuriti in the lant effect the fragmentan npI' c . Th e areas for f'tllure re'e3J'ch are ju t a fewtI,at c uLdfurther the relev,lI1ce of experimental anal gsto pili' atomagl11atic volcani m.

SEE ALSO THE FOLLOWI G ARTICLES

Magmatic Fragmentation· Phrcatopliruan 'ruptions·urt cyan and Rchlted Phreatomagmati Eruptions· eli"

ne' and Tuff Ring •. ulca11ian Erupci 11

FURTHER READI G

Buet01er, R. t and Zim:mowski B. (199 ). Phy ic ofthermohy­draulic expl ions. Phys. Rev. E. 57(5) -726-5729.

PHREATOMAGNI,\TI FRA 1\'1' TAT10l\' 445

19ntc . ,:1nd igurgeir s n, T (1973). Dynamic mixingof W:1ter and IllV:1. !Vfltlm: 244, 552-555.

Heikcn, '., and ohlel7.. K. (1985).' olcanic h." Uni¥ r­i f :lliforn.i:l Pre Berkeley,

Lorenz., . L9 6). n the growth of maars and diatremes andi . rcle ance to the forma Don of ruff rings. BllII. Volcano!.4 _6-- 4.

herid:m 't. ., and ohletz., K H. (19 3). Hydro olcanism:Basic con 'dcrati ns and review.]. Vokmlol. Geutherm. Res.17 1-29.ohletz., K H.. md Heiken, . (1991). "Volcanology aTld Ceo­rhn71/fl1 Ellergy. nive i~' of California Press B rkelcy,

ohle H. and M ueen, R. G. (19 4). In • hdies inGeopbyri" '. R. Boyd Jr. cd). ationa! Academy Pr ,\ a hingt nO.

W hi tz., K. H. 19 3). 1ech.anisms of hydro leanie pyr -

cia t f< mlao n: r:illl ize canning el etron micro copyand xperimenmJ results.]' Volennol. Geotbenll. Res. 17,

1-63.ohlen K I . (19 6). xpl ive nl:lgma-watcr interactions:Th nn dyn:uni e.-yplo ion mechanisms and field studies.8u//. okanol. 48 245-264.ohletz., H. I L ueen R. G. and [omssey, . (1995).Pag - 1 111 Intense Mliltiphose Interactions" (T.The fano and rna ds.), Proceedings of

F)Japan 0 P )Joint emiuar anea BarbaTll June-L L995.

Zim:m \\ . B., Buettner R., Lorenz, \. and Haefele H.- .J99). .ragm n cion of ba alcie mdt in the course of

expl 've Icanisrn.J. Ceopbys. Res. 102 03 14.Zimano\ ski B. Bue er R. and Lorenz, \. 199). Pre­

mixing of magma and water in IFCI ~"erimen . Bull.olemo/. 5 ... 1-4 ~.