Growth mechanism of snowball garnets from the Lukmanier Passarea (Central Alps, Switzerland): a combined lCT ⁄EPMA ⁄EBSDstudy

Martin Robyr,1 Pierre Vonlanthen,2 Lukas P. Baumgartner1 and Bernard Grobety2

1Institute of Mineralogy and Geochemistry, University of Lausanne, CH-1015 Lausanne, Switzerland; 2Department of Geosciences, University

of Fribourg, Chemin du Musee 6, CH-1700 Fribourg, Switzerland

Introduction

Snowball garnets, also referred to asspiral-shaped garnets, are character-ized by the presence of sigmoidalinclusion trails with a relative rotationangle exceeding 90� (Williams andJiang, 1999). The formation of snow-ball garnets is explained by two con-troversial models: one argues forporphyroblast growth with rotationin respect to the foliation in a singletectonic phase of non-coaxial flow(e.g. Schoneveld, 1977; Passchieret al., 1992), whereas the other arguesfor porphyroblast growth withoutrotation, involving overgrowth ofsuccessive generations of nearorthogonal foliations (e.g. Bell et al.,1992). So far, most efforts to assessthese two models have focused ongeometrical proofs. However, thisapproach has failed to resolve theissue completely as most geometriesobserved in snowball garnets can beexplained by both the rotational andthe non-rotational models (e.g.Johnson, 1993; Stallard, 2003).An alternative way of deciphering

the development of snowball garnetporphyroblasts consists of combining

chemical zoning and crystallographicorientation analysis with geometricalarguments. In a recent paper, Ikedaet al. (2002) used a combinedapproach to account for the forma-tion of snowball garnets. Based on thechemical zoning and the monocrystal-line nature of the investigated garnets,they concluded that growth and rota-tion of garnet occur simultaneouslyduring a single metamorphic event. Inlight of new evidence provided bymicro-computed X-ray tomography(lCT), electron probe microanalysis(EPMA) and electron backscattereddiffraction (EBSD) techniques, we pre-sent an alternative model to accountfor the growth of snowball garnets.

Geological setting

The samples were collected in theLukmanier Pass area, Central Alps,Switzerland (Fig. 1). This region com-prises two major tectonic units: theGotthard Massif to the north, partlyoverlain by a Mesozoic sedimentarycover, and the Lukmanier Massif tothe south. Previous investigations (e.g.Chadwick, 1968) show that theLukmanier area has a complex tec-tonic history involving three mainphases of deformation. All of themare related to southwards and up-wards tectonic movements resultingfrom the Tertiary continent–continentcollision between the Eurasian andApulian plates.

The two initial phases of deforma-tion (D1 and D2) correspond to acrustal thickening event related to top-to-the-N directed movements whichmost probably reflect the southwardunderthrusting of the Mesozoic sedi-mentary cover beneath the frontalpart of the Apulian plate. Whereasthe first deformation (D1) was respon-sible for isoclinal folding and imbri-cate stacking of basement andsedimentary cover units, the D2 phasegenerated the axial planar surface ofthe large isoclinal folds observed inthe Lukmanier Pass area. Intensive

ABSTRACT

For two decades, considerable efforts have been made toexplain the formation of snowball garnets by either therotational or non-rotational models. On the basis of morpho-logical, chemical and crystallographic evidence, this paperpresents new data on snowball garnets showing that theformation of these microstructures can be explained by thecombination of the two previously proposed mechanismsoperating consecutively during garnet growth. The crystalliza-tion sequence of garnet revealed by Mn contouring and thedistribution of crystallographic orientations within the spiral

indicate that the final stages of garnet growth are controlled bypost-kinematic crystallization. However, some microstructuralarguments plead for a rotational contribution during the firststages of growth. In this view, the overall spiral geometry isthought to overestimate the true amount of rotation experi-enced by the garnets. Results also reveal the existence ofcomplex snowball garnets consisting of several grains formedfrom distinct nucleation sites.

Terra Nova, 19, 240–244, 2007

Correspondence: Dr Martin Robyr, Geo-

logical Science Department, The University

of Texas at Austin, 1 University Station

C1100, Austin, TX 78712-0254, USA. Tel.:

+1 512 471 9425; fax: +1 512 471 8054;

e-mail: mrobyr@jsg.utexas.edu

Lukmanier Massif

Gotthard Massif

Stgir series

Pizzo di Cadreigh

Scopi

N

2 km

Switzerland

Studied area

Lukmanier Pass

Brönich

Sta

Maria

Lake

Mesozoic sedi- mentary cover

Fig. 1 Simplified geological map of theLukmanier Pass area. The sampling loca-tion (708.163 ⁄156.045, Swiss nationalgrid coordinates) is indicated by a star.

240 � 2007 Blackwell Publishing Ltd

doi: 10.1111/j.1365-3121.2007.00741.x

SE-directed back-folding and back-thrusting (D3) subsequently causedthe upward movements of crustalmaterial and the exhumation of theLukmanier metamorphic rocks.The occurrence of snowball garnets

is restricted to specific levels of theLiassic Stgir series (cover of theGotthard Massif) (Chadwick, 1968;Fox, 1975), which are characterizedby thin alternations of quartz andmica-rich layers. The main structuralelement in those garnet-bearing levelsconsists of a N-dipping penetrativecrenulation cleavage (S2) (Fig. 2a),associated with a monotonous N–Smineral lineation. Shear indicatorsclearly show a top-to-the-N directedsense of shear for S2. The relationshipbetween crystallization and deforma-tion indicates that the main stage ofgarnet growth occurred contempora-neously with the S2 formation, i.e.during the D2 phase. The overprint-ing of S2 over the earlier schistosity(S1) resulted in the formation ofclosely spaced microlithons display-ing a subhorizontal E–W crenulationlineation parallel to the rotation axisof the snowball garnets. Both S1 andS2 foliations were subsequently foldedto form asymmetric, centimetre tokilometre-scale folds (F3) verging tothe SW. A slightly NE-dipping crenu-lation cleavage (S3) is locally visiblein the hinges of the F3 folds, but nopenetrative schistosity is reported.Occasionally, D3 produced shearbands (SB3) with top-to-the-S direc-ted movements (Fig. 2b).The studied samples contain the

mineral assemblage garnet + musco-vite + biotite + quartz + plagioclasewith minor epidote, chlorite andilmenite. Although the garnet-formingreaction has not yet been fully esta-blished, the systematic occurrence ofgarnet crystals along mica-rich layersstrongly suggests mica as reactant.Regional mapping of the mineral dis-tribution throughout the LukmanierPass indicates a metamorphic fieldgradient characterized by peak condi-tions increasing gradually from thechloritoid zone to the north of theSanta Maria Lake, to the staurolite–kyanite zone in the Bronich area tothe south (Fox, 1975). Thermobaro-metric analysis of metapelites indi-cates peak conditions evolving from410 �C for the chloritoid zone (Rahnet al., 2002) to 550 �C ⁄5.5 kbar for

the staurolite–kyanite zone (Engi et al.,1995).

Analytical procedures

Three-dimensional imaging was ob-tained by lCT using a high-resolutionSkyscan-1072 system (SKYSCAN,Kontich, Belgium) with a conical100 kV X-ray source. A Cameca SX50 electron microprobe (CAMECA,Gennevilliers Cedex, France) (beamcurrent of 150 nA at 25 kV) was usedfor the mapping of chemical elementsthrough wavelength dispersive X-rayspectroscopy (WDS). Crystallograph-ic orientations were collected byEBSD using a Philips FEI XL30SFEG Sirion scanning electron micro-scope (Philips, Eindhoven, Nether-lands) (probe current of 20 nA foran acceleration voltage of 30 kV).

Three-dimensional geometry

Micro-computed X-ray tomographyimages reveal that the geometry ofmost garnets is characterized by a coreregion, around which two spiral armsare wound over more than 300�.Because of their very elongated shapeparallel to the apparent rotation axis,the studied garnets look more likerolled cylinders than snowballs. Thecore region and the base of the spiralsare commonly connected with theadjacent arms by short, curved garnetbridges (Fig. 3a). According toSchoneveld (1977), bridges are createdby thin mica-rich layers progressivelycaptured from the surrounding matrix

AS3

AS3

AS2

AS2

S2

S2

S2

S2

SB3

2 mm

S1

S1

SB3

(b)

(a)

N S

S2

Fig. 2 Photomicrograph (a) and sketch(b) of the same area illustrating therelationship between the snowball gar-nets and the foliations S1, S2 and S3. AS,axial surface; SB, shear bands.

1 mm

S1

S1

5 mm

(c)(b)

(a)

Fig. 3 (a) lCT image of sampleLuk_02_5 illustrating the typical spiralmorphology of snowball garnets. Thecore region of the spiral is connected tothe arms by thin curved bridges (blackarrows). (b) lCT view of three verticallyaligned garnets. The two porphyroblastson top are connected to each other by athick garnet appendage (white arrow).(c) Selected two-dimensional sections(normal to the cylinder axis) based onlCT data showing trails of garnet grains(in black) mimicking the foliation Se(dashed line) of the host rock.

Terra Nova, Vol 19, No. 4, 240–244 M. Robyr et al. • Growth mechanism of snowball garnets

.............................................................................................................................................................

� 2007 Blackwell Publishing Ltd 241

and dragged towards the centre as thespiral forms. Replacement of mica bygarnet is commonly believed to freezethe bridge geometry.Neighbouring spiral garnets are

occasionally linked to each other byappendages (Fig. 3b), consisting ofsmall aligned garnet grains. Selectedtwo-dimensional sections of tomo-graphic images reveal that the trailsleft by these grains mimic the S1foliation (Fig. 3c). Optical microscopyobservations indicate that most ofthese grains occur within mica-richlayers. The correspondence in geo-metry between the tortuous andcrenulated external foliation (S1) andthe alignment of garnet grains repla-cing mica suggests that the nucleationand growth of garnet continued afterD2 had stopped.

Chemical zoning

Compositional zoning is a recurrentcharacteristic of garnet porphyro-blasts, commonly interpreted as a pri-mary growth feature (e.g. Spear, 1993).Because of its very limited partitioningwith the other phases present in peliticschists, Mn is generally considered asthe element best reflecting the succes-sive steps of garnet growth.Assuming alocal chemical equilibrium along thegarnet–matrix boundary and no post-growth intracrystalline diffusion, thegradual decrease in Mn concentrationfrom core to rim is frequently used as aproxy of time reflecting garnet growthhistory (e.g. Kretz, 1973; Carlson,1989, 1991; Spear and Daniel, 1998).X-ray compositional maps show a

broadly concentric zoning of mostmajor elements, characterized by agradual decrease of Mn (Fig. 4a andc) and Ca from core to rim, counter-balanced by a simultaneous increaseof Fe and Mg. The strong correlationin the zoning of Mn, Ca, Fe and Mgsuggests that these elements achievedlocal equilibrium during garnetgrowth, thus justifying the use of Mnas a time marker. Although the vari-ation in Mn concentration shows aroughly concentric pattern, secondaryMn maxima do exist, e.g. in the leftspiral end in sample Luk_02_1(Fig. 4a), as well as, to some extent,in the upwards-oriented spiral termi-nation in sample Luk_02_7 (Fig. 4c).The gradients on contoured Mn

maps of sample Luk_02_1 (Fig. 5)

indicate not only a decrease in Mnconcentration from core to rim in thecentre and along the spiral curvature,but also across two bridges connectingthe core to the arms. The patternsobserved after considering Mn con-centration levels as time lines suggestthat the areas of the arms located inthe continuation of the bridges crys-tallized early in the microstructuredevelopment, before being reachedby the crystallization front advancingalong the spiral curvature. In thisgrowth scheme, the bridges served asshortcuts, making the crystallizationpath from the core to the arms signi-ficantly faster. Subsequent growth,either from the bridges or the secon-dary nucleation sites, led to theformation of the final garnetmicrostructure.

Crystallographic orientation

Electron backscattered diffraction isused to identify possible changes incrystallographic orientation within thegarnet spirals. By assuming a continu-ous incorporation of atoms onto thepre-existing garnet structure, a singlecrystallographic orientation would beexpected for the whole spiral, as wasobserved by Ikeda et al. (2002) in theirX-ray-based texture analysis of snow-ball garnets of different origins. OurEBSD maps, however, reveal that thesnowball garnets from the LukmanierPass are composed of several grainswith different, distinct crystallo-graphic orientations. The core regionand the base of the spiral arms aregenerally formed by one single, largegrain, whereas the arm terminations

1 mm

(a)

(b)

(c)

(d)

1 mm

Fig. 4 WDS X-ray maps of Mn (top, concentration decreasing from red to blue) andEBSD crystallographic orientation maps (bottom) for samples Luk_02_1 (a and b)and Luk_02_7 (c and d). Secondary Mn maxima (arrows) are centred in grains withcrystallographic orientations different from the one of the core region. In sampleLuk_02_1, the primary core orientation is maintained in two segments of the spiralarms connected to the centre by garnet bridges (b, light green), along which a smoothdecrease in Mn concentration is observed (a).

Growth mechanism of snowball garnets • M. Robyr et al. Terra Nova, Vol 19, No. 4, 240–244

.............................................................................................................................................................

242 � 2007 Blackwell Publishing Ltd

are made of smaller and often box-shaped crystals separated by straightgrain boundaries (Fig. 4b and d). Theangular misorientation between thosegrains is typically a few tens ofdegrees, indicating that they are notsubgrains formed by deformation ofthe arms. Some of the box-shapedgrains contain secondary Mn maxima.Both the difference in crystallographicorientation and the occurrence ofadditional Mn maxima point togrowth from new nucleation sites.Among the end grains, however, someof them do not show any anomalous

zoning. This may be explained by thefact that the thin sections used forchemical mapping crosscut thesegrains at some distance from theirnucleation centre.Garnet grains with different lattice

orientations totally fill the spiralterminations in sample Luk_02_7(Fig. 4d). In sample Luk_02_1(Fig. 4b), however, the primary coreorientation is maintained in an inter-stitial segment of the upper arm and inthe final segment of the lower arm.Both portions are connected to thecore by a thin, curved garnet bridge.As previously discussed, based onchemical Mn zoning patterns, thesebridges were interpreted as preferen-tial crystallization paths between thecore and the arms, suggesting anearlier crystallization of garnet acrossthe bridges. The correspondence incrystallographic orientation betweentwo regions of the spiral separated bygrains with different lattice orienta-tions further supports the role ofcrystallization shortcuts played bymica bridges. For such a crystalliza-tion path to occur, the position andfinal curvature of the spiral arms hadto be already acquired before garnetcrystallized. Therefore, we suggestthat significant parts of the overallgeometry were completed relativelyearly in the microstructure develop-ment and that the replacement of micaby garnet in the last portions of thespiral arms was mostly post-defor-mational.

Discussion and conclusions

Combined chemical and crystallo-graphic analyses provide severalpieces of evidence showing that thespiral-shaped overall geometry was

already achieved prior to the laststages of garnet crystallization. Thisstrongly supports a final post-kine-matic growth of garnet, mainly byreplacement of mica.The mechanism responsible for the

first stages of snowball garnet forma-tion is more difficult to assess; how-ever, assumptions can be made on thebasis of microstructural observations.During the crustal thickening event(D2), garnets grew on the overturnedlimbs of microlithons (Fig. 6) and arealigned in a direction correspondingto the crenulation lineation. This indi-cates that the first step of garnetcrystallization post-dated the begin-ning of the microlithon formation andoccurred when the D2 deformationwas still in progress. The growth ofsnowball garnets on the overturnedlimb of microlithons during a singledeformation phase is best explainedthrough a single and continuous rota-tion of garnet grains rather thanthrough an incremental superimposi-tion of foliations resulting fromnumerous changes in the stress field.Our model suggests that a rotation of180� may be sufficient to produce aspiral-shaped geometry mimicking anamount of rotation of approximately300� (Fig. 7). This implies that thefinal geometry of snowball garnetsgreatly overestimates the actual extentof rotation. This assumption does notdispute the shear sense in the foliationplane as it is determined by the senseof rotation of the garnet during itsinitial growth stage.Our data support the idea that the

growth of the snowball garnets fromthe Lukmanier area cannot beexplained by either of the two classicalmodels considered separately. Incontrast, this study illustrates that

(c)

(b)

(a)

1 mm

Fig. 5 Contoured Mn maps retracingthe growth history of sample Luk_02_1.Arrows indicate the direction of growth.

N

1 cm

S

S2

S1

Fig. 6 Sketch cross-section perpendicular to the crenulation axis showing thedistribution of snowball garnets relative to S1 and S2. Garnet grains are locatedpreferentially on the inverse limbs of microlithons.

Terra Nova, Vol 19, No. 4, 240–244 M. Robyr et al. • Growth mechanism of snowball garnets

.............................................................................................................................................................

� 2007 Blackwell Publishing Ltd 243

snowball garnets may undergo acomplex growth history in whichboth end-member mechanisms canoperate consecutively. Additionalchemical, structural and crystallo-graphic analyses on a larger samplepopulation are required to betterunderstand the mechanism of snow-ball garnet formation, and to deter-mine if our observations are valid on alarger scale.

Acknowledgements

We particularly wish to thank J.-L. Epardand J.-C. Vannay for the fruitful discus-sions we had with them and for helping usout in the field. We also thank two anony-mous reviewers for their very helpful com-ments. This study was supported by theSwiss National Science Foundation(FNRS-grants 2100-066996 and 200020-103865).

References

Bell, T.H., Johnson, S.E., Davis, B., Forde,A., Hayward, N. and Wilkins, C., 1992.Porphyroblast inclusion-trail orientationdata; eppure non son girate! J. Meta-morph. Geol., 10, 295–307.

Carlson, W.D., 1989. The significance ofintergranular diffusion to the mechanismand kinetics of porphyroblast crystal-lization. Contrib. Mineral. Petrol., 103,1–24.

Carlson, W.D., 1991. Competitive diffu-sion-controlled growth of porphyro-blasts. Mineral. Mag., 55, 317–330.

Chadwick, B., 1968. Deformation andmetamorphism in the Lukmanier region,central Switzerland. Geol. Soc. Am. Bull.,79, 1123–1149.

Engi, M., Todd, C.S. and Schmatz, D.R.,1995. Tertiary metamorphic conditionsin the eastern Lepontine Alps. Schweiz.Mineral. Petrogr. Mitt., 75, 347–369.

Fox, J.S., 1975. Three-dimensionalisograds from the Lukmanier Pass,

Switzerland, and their tectonic signifi-cance. Geol. Mag., 112, 547–564.

Ikeda, T., Shimobayashi, N., Wallis, S.R.and Tsuchiyama, A., 2002. Crystallo-graphic orientation, chemical composi-tion and three-dimensional geometry ofsigmoidal garnet: evidence for rotation.J. Struct. Geol., 24, 1633–1646.

Johnson, S.E., 1993. Testing models for thedevelopment of spiral-shaped inclusiontrails in garnet porphyroblasts: to rotateor not to rotate, that is the question.J. Metamorph. Geol., 11, 635–659.

Kretz, R., 1973. Kinetics of the crystal-lization of garnet at two localities nearYellowknife. Can. Mineral., 12, 1–20.

Passchier, C.W., Trouw, R.A.J., Zwart,H.J. and Vissers, R.L.M., 1992. Por-phyroblast rotation; eppur si muove?J. Metamorph. Geol., 10, 283–294.

Rahn, M.K., Steinmann, M. and Frey, M.,2002. Chloritoid composition and for-mation in the Central Alps: a compar-ison between Penninic and Helveticoccurrences. Schweiz. Mineral. Petrogr.Mitt., 82, 409–426.

Schoneveld, C., 1977. A study of sometypical inclusion patterns in stronglyparacrystalline-rotated garnets. Tecton-ophysics, 39, 453–471.

Spear, F.S., 1993. Metamorphic PhaseEquilibria and Pressure–Temperature–Time Paths, pp 582. MineralogicalSociety of America Monograph.

Spear, F.S. and Daniel, C.G., 1998. Three-dimensional imaging of garnet porphy-roblast sizes and chemical zoning;nucleation and growth history in thegarnet zone. Geol. Mater. Res., 1, 1–43.

Stallard, A., 2003. Comment on ‘‘Crystal-lographic orientation, chemical compo-sition and three-dimensional geometry ofsigmoidal garnet: evidence for rotation’’by T. Ikeda, N. Shimobayashi, S. Wallisand A. Tsuchiyama. J. Struct. Geol., 25,1337–1339.

Williams, P.F. and Jiang, D., 1999. Rota-ting garnets. J. Metamorph. Geol., 17,367–378.

Received 22 August 2006; revised versionaccepted 22 December 2006

0° 75° 120° 180°

S2

(a)

(b)

S2

N S

Fig. 7 (a) Growth model proposed for the studied snowball garnets. Starting from anucleus on the inverse limb of the microlithon, a 180� rotation is sufficient to accountfor an apparent rotation of about 300�. (b) Idealized sketch showing the alignment ofgarnet grains in a crenulated rock. Syn-rotational growth is shown in black, post-deformation growth in grey.

Growth mechanism of snowball garnets • M. Robyr et al. Terra Nova, Vol 19, No. 4, 240–244

.............................................................................................................................................................

244 � 2007 Blackwell Publishing Ltd