Journal of Petrology Advance Access originally published online on December 26, 2008
Journal of Petrology 2009 50(1):97-102; doi:10.1093/petrology/egn074
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Response to Comment by McBirney, Boudreau and Marsh
1Department Of Earth Sciences, University Of Cambridge, Downing Street, Cambridge CB2 3EQ, UK
2Department Of Geosciences, University Of Massachusetts, 611 North Pleasant Street, Amherst, MA 01003-9297, USA
3Department Of Earth Sciences, University Of Aarhus, HØEgh-Guldbergs Gade 2, Dk-8000, Aarhus C, Denmark
RECEIVED NOVEMBER 20, 2008; ACCEPTED DECEMBER 5, 2008
 |
INTRODUCTION |
|---|
TOP
INTRODUCTION
SUB-SOLIDUS HISTORYThe concept of variable textural maturity at grain junctions in igneous rocks was first introduced by Holness et al. (2005
). The papers of Holness et al. (2007a, 2007b) demonstrate how this concept can be applied to cumulates, and present a hypothesis to explain the observation of excursions in textural maturity, either localized, or step-wise, within a cumulate pile. The comments by McBirney et al. refer to difficulties they have in accepting our hypothesis, and we welcome the opportunity to clarify some confusions in Holness et al. (2007a, 2007b). Here we identify the issues to be addressed with separate headings. |
SUB-SOLIDUS HISTORY |
|---|
The first point raised by McBirney et al. was prompted, perhaps, by our unqualified use of the term ‘sub-solidus’: here we explicitly address the sense in which we used it. The studies of Holness et al. (2007a
, 2007b) are based on variations of the clinopyroxene–plagioclase–plagioclase dihedral angle, or 
cpp. During solidification of basaltic liquid, liquid-filled pores in a plagioclase crystal framework are gradually filled by clinopyroxene. The point at which this occurs for a particular pore corner defines the onset of the ‘sub-solidus’ for this junction: it is therefore applied locally. It is very important to remember that even when the melt at one junction has been replaced by pyroxene there may still be some liquid present at other plagioclase–plagioclase junctions: the rock is not necessarily completely solid.
The infilling pyroxene inherits the pore shape, including the solid–solid–melt angle (the dihedral angle) subtended at pore corners (Holness et al., 2005, 2007c). The initial cpp is therefore a faithful copy of the melt–plagioclase–plagioclase angle and as such is far from solid-state textural equilibrium. Given the opportunity to anneal it will change by a diffusive process. The amount by which it changes will be a function of how long the junction remains above the closure temperature for multi-component grain boundary diffusion, or ‘how hot for how long’. Clearly the (sub-solidus) textural maturation clock cannot start ticking until all the melt at the local plagioclase–plagioclase junction has been replaced by pyroxene. ‘How hot for how long’ then becomes ‘the time-integrated history in the sub-solidus’ (Holness et al., 2007a, 2007b).
The importance of the relative timing of the creation of sub-solidus three-grain junctions was very clearly demonstrated by Holness et al. (2005), who showed that the earlier-formed clinopyroxene–olivine–olivine junctions in the centre of a clinopyroxene oikocryst have higher dihedral angles than the later-formed junctions on the oikocryst margin: the textural maturation clock started ticking earlier in the centre, while interstitial liquid was still adjacent to olivine–olivine junctions only a few millimeters away. Similar differences in angle are also seen in clinopyroxene oikocrysts enclosing plagioclase (Holness, 2005). These observations clearly indicate a time sequence of local sub-solidus closure events.
|
DIRECT EVIDENCE FOR MAGMA RECHARGE EVENTS AFFECTING TEXTURAL MATURITY |
|---|
McBirney et al. find it difficult to believe that
cpp can be affected by processes occurring in the bulk magma adjacent to the porous cumulate. Holness (2005, 2007) demonstrated very clearly that the cpp in allivalites (the local field term for plagioclase-rich troctolitic cumulates) of the Rum Eastern Layered Intrusion increases significantly immediately beneath overlying peridotite layers and ascribed this to the heat pulse associated with replenishment of the magma chamber by picritic melt. In Holness et al. (2007a), a similar increase in cpp is associated with a step-change in 87Sr/86Sr in the Rum Unit 14 allivalite, which can be explained by an influx of fresh, basaltic, magma into the chamber (Renner & Palacz, 1987). We consider the association of increased cpp both with chamber-wide changes in cumulate mineralogy to more primitive compositions and with more contaminated isotopic compositions to be strong evidence that the time-integrated thermal history of clinopyroxene–plagioclase–plagioclase junctions in the crystal mush on the chamber floor is indeed affected by the arrival of fresh magma in the chamber. Fresh, hot, magma placed adjacent to the crystal mush keeps the clinopyroxene–plagioclase–plagioclase junction hotter for longer.
A similar line of reasoning applies to the explanation of the spikes in textural maturity observed in the basal section of the Skaergaard Intrusion (Hidden Zone), which we interpret as evidence for pulsed filling of the magma chamber (Holness et al., 2007a). In contrast, the upper levels of the Skaergaard Intrusion formed during steady evolution in a closed system, in which the heat effects are not associated with fresh injection of magma but instead with the sequential evolution of the magma through a series of new cotectic equilibria involving new crystal phases.
|
THE OBSERVED LINK BETWEEN DIHEDRAL ANGLE AND LIQUIDUS ASSEMBLAGE |
|---|
McBirney et al. also find it hard to believe that changes in
cpp can record changes in the liquidus assemblage, and indeed the bulk of their comments refer to this difficulty. We wish to stress, however, that the starting point for Holness et al. (2007a, 2007b) is purely observational. Holness (2005, 2007) and Holness et al. (2007a) find that the dihedral angle in the troctolitic cumulates from the Eastern Layered Intrusion of the Rum Central Complex is consistently lower than those in the associated olivine gabbroic cumulates. The open nature of the Rum magma chamber means that successive allivalites are sometimes troctolitic and sometimes olivine-gabbroic—in each case the dihedral angle faithfully switches from low values (troctolite) to higher values (olivine gabbro). The relationship is even more clearly shown in the closed system provided by the Skaergaard Intrusion: the change from troctolitic cumulates of LZa to the gabbroic cumulates of LZb corresponds to a step-change in dihedral angle (Holness et al., 2007a). Subsequent work on the Skaergaard Layered Series, reported in Holness et al. (2007b), demonstrates that similar step-changes in dihedral angle occur at the arrival of Fe–Ti oxides as a cumulus phase, and at the arrival of cumulus apatite. Preliminary work on the Marginal Border Series of the Skaergaard Intrusion shows identical step-changes associated with zonal boundaries. We find inescapable the conclusion that these changes in dihedral angle are related to changes in the liquidus assemblage. We summarize the mechanism by which this happens in our further remarks and Conclusions, below. |
THE METASOMATISM HYPOTHESIS |
|---|
McBirney et al., although dismissing the link between changing liquidus assemblage and
cpp, do not offer an alternative hypothesis other than suggesting that metasomatism may play an (unspecified) role in the case of the Skaergaard LZa/b event; they are silent on the subject of the step-changes in cpp at oxide-in and apatite-in and do not comment on the data from Rum. In contrast, Holness (2007) and Holness et al. (2007d) demonstrate clearly how metasomatism may alter cpp. A detailed study of the Unit 10 allivalite on Rum, previously shown by Tait (1985) to have undergone a localized metasomatic event, shows that cpp is reduced by the infiltration of late-stage, broadly chemically equilibrated liquids (Holness, 2007). Such infiltration is associated with a distinctive textural and geochemical signature. Metasomatism of the Unit 9 allivalite, following pervasive infiltration of a primitive, chemically reactive liquid, results in a characteristic sigmoidal cpp record through the metasomatized allivalite, together with metre-scale removal of clinopyroxene from the lower reaches of the allivalite and its reprecipitation in overlying horizons to form an augite-enriched olivine gabbro (Holness et al., 2007d).
The metasomatic features referred to by McBirney et al. all occur in the Skaergaard Intrusion. The sets of samples collected by Karen Bollingberg in 1995 and Christian Tegner (inter alia) in 2000, which span the entire stratigraphy of the exposed Layered Series, were carefully selected from average gabbro, avoiding any layers with modal compositions far from the cotectic. None of the samples from either of the two drill cores (the 1966 Cambridge core and the 90-22 core) show any unusual modal compositions. We do not see any textural evidence in our Skaergaard samples that would tend to compromise our observations: the thin sections all have textures entirely consistent with those observed in glassy crystalline nodules (Holness et al., 2005, 2007c; Holness, 2006), and not with any degree of later modification (e.g. Holness, 2007).
The internal consistency of many dozens of data points between our discontinuities, some collected along strike from widely spaced parts of the intrusion, would require that they were all metasomatically modified to the same extent (a reasonably improbable supposition) or have not been seriously altered at all. We suggest that the effects cited in the list of references given by McBirney et al. are relatively local in occurrence and that neither the three step-changes seen in the Skaergaard cpp data nor the bimodal cpp data from the Rum allivalites are recording previously undescribed chamber-wide metasomatic events. We are left with the conclusion that the step-changes are a primary result of changes in the liquidus assemblage.
|
CONVECTION CONFUSION |
|---|
McBirney et al. refer to the role of convection in heat loss from solidifying magma chambers. The ‘onset of chamber-wide convection’ quoted by McBirney et al. refers to events at Skaergaard (rather than in Rum, as erroneously stated by McBirney et al.) in the basal thermal boundary layer discussed by Maaløe (1976
) after a fresh magma input, and quite logically distinguishes chamber-wide convection as being distinct from the plausibly more chaotic state preceding. The shift at Skaergaard from a convective to a conductive regime cited by McBirney et al. refers to the later stages of Skaergaard crystallization in UZb where Jang & Naslund (2001) suggested cessation of chamber-wide convection. In other words, this represents the dying out of wholesale magmatic convection owing to solidification, and the onset of a predominantly conductive regime. The critical comments are unrelated to the issues being discussed. The statement of McBirney et al. that it ‘seems [that we] assume that the rate of heat loss is more rapid if the magma is convecting’ appears not to be relevant to the situation under discussion above, but contains a mistake of general interest. The advective transport of heat on an adiabat in a hot fluid medium is much more efficient than a purely conductive transport, and so in turn convection brings the hottest possible contrast to the main thermal boundary layer. The flux of heat into the surroundings varies positively with the thermal contrast, and to the extent that the maximum heat contrast is maintained, the rate of heat loss will indeed be more rapid if the magma is convecting.
|
ENTHALPY EFFECTS |
|---|
In Holness et al. (2007a
, 2007b) we attempt to explain the link between textural maturity and liquidus assemblage by examining changes in the contributions to the total enthalpy budget. McBirney et al. suggest that our model is based on a mistaken interpretation of Wyllie (1963). Wyllie did indeed describe a change in the amount of crystallization per degree of temperature decline on the addition of a second phase to the liquidus, and we stated this explicitly on p. 2369 of Holness et al. (2007b), where we also invoked ‘a regular continuum of heat loss’ to the surroundings, in agreement with McBirney et al. Where we depart from McBirney et al. is in their following paragraph where they state ‘given a constant rate of heat loss and minerals with similar heats of crystallization, the total amount of crystallization remains nearly constant regardless of the number of minerals on the liquidus’. This statement can be true only if the liquidus slope also remains constant regardless of the number of minerals on the liquidus. Wyllie (1963), together with numerous other studies, shows that this is not the case.
We regret that we did not clearly define our terminology in Holness et al. (2007a): this oversight has led to confusion amongst authors, reviewers and readers, and was only remedied in an extensive passage in Holness et al. (2007b) at page 2369, which sets out very clearly the concept of fractional latent heat and the way in which this must change during progressive solidification. We agree with McBirney et al. that the principles involved are fundamental, but their illustrations are not helpful in understanding these principles. Their graphs show the total mass crystallized as a function of temperature and of the bulk enthalpy. However, as carefully explained in Holness et al. (2007b), it is the changes in fractional latent heat that affect the time-integrated thermal history of each clinopyroxene–plagioclase–plagioclase junction. Here we illustrate this principle in a more informative plot, Fig. 1.
|
|
FRACTIONAL LATENT HEAT |
|---|
Figure 1 combines a variant of Ghiorso's (1997
) fig. 5 with a new diagram of the fractional latent heat in a mid-ocean ridge basalt (MORB) composition undergoing fractional crystallization. This figure shows the important relationship of the latent heat of fusion to the temperature of crystallization. It is clear that each new phase brings with it a new burden of heat to be removed. In consequence, the crystal productivity increases and the specific cooling rate decreases, as defined and described on p. 2369 of Holness et al. (2007b). In Fig. 1a, the peaks of latent heat are particularly sharp and large for olivine and augite. A lesser peak reflects the appearance of titanomagnetite, followed by a small peak for apatite. From these alone it might seem that magnetite and apatite would have only minor effects on the cooling history, but that would be wrong.
Figure 1b is a plot of the fractional latent heat as a function of progressive solidification. It shows the discontinuous jumps described from theory on p. 2369 of Holness et al. (2007b); these clearly show the overwhelming importance of the relative contributions to the enthalpy budget of latent heat and sensible heat. A fractional latent heat path is intrinsically discontinuous in fractional crystallization because the instantaneous solid composition jumps discontinuously (the ‘rock hop’ of Morse, 1994) at the addition of each new crystalline phase to the fractionating assemblage. The continuous, cotectic liquid path then changes course in composition space to run away from the new solid assemblage.
In detail, the peak for titanomagnetite is sharp, rising abruptly over an interval of only 4 degrees, but it is preceded by an unexpected premonitory increase over a range of about 25 degrees. This effect of a more gradual approach to a latent heat peak may help to explain the intermittent presence of cumulus magnetite grains in the underlying cumulates at Skaergaard, leading to a smeared arrival of the new phase. The peak height for titanomagnetite is slightly smaller than for augite, whereas the peak jump for apatite is very sharp and respectable in height. The slow and continuous rise in fractional latent heat after apatite-in reflects a continual steady approach to a maximum without further phase changes. As predicted, the function runs to 1·0 at the end point of the fractionation sequence.
As mentioned in Holness et al. (2007b), any modal overshoot at the arrival of a new phase would tend to amplify the intrinsic effects of the latent heat jumps shown in Fig. 1b. We cannot phrase it better than the felicitous comment of Mark Ghiorso: ‘when a new phase appears on the liquidus, nature sits and enjoys the fact for a bit while the latent heat conducts itself away’ (e-mail to S. A. Morse, 15 September 2008).
The driving force for textural maturation is directly proportional to the angular difference between the initial and final states. It is also true that the ‘hotter for longer’ principle is proportional to the magnitude of the fractional latent heat pulse. Hence we expect to find the chance for textural maturation greatest at Aug+, less at Mt+, and least at Ap+. This sequence corresponds to our observations.
McBirney et al. do not address the plot of latent heat against temperature for MORB as shown in our Fig. 1a, even after we called to their attention the Ghiorso (1997) version in an earlier preview of this reply. Without this dramatic representation, their calculations entirely miss the point of our argument. The extension from our Fig. 1a to Fig. 1b is simply a graphical rendition of our theoretical development in Holness et al. (2007b).
|
SOLIDIFICATION FRONTS |
|---|
Solidification fronts related to accumulation fronts have been with us in the study of layered intrusions since Hess (1939
), Wager et al. (1960) and Morse (1982), among many others. Historically, the term solidification front (or interface in Morse, 1986) is the notional interface between solid cumulates and their adjacent magma, normally a mushy zone of some finite thickness, although Marsh has adopted the term to signify the outermost surface of such a mushy zone. His large mushy zone thicknesses, perhaps applicable to sills, are incompatible with the slow mean accumulation rate of Skaergaard cumulates (
2 cm/year by our reckoning from literature estimates) and the evidence for considerable adcumulus growth based on plagioclase zoning (Toplis et al., 2008).
The solidification front is the locus of the maximum temperature in a crystallizing layered intrusion (Irvine, 1970, fig. 17), a ‘latent heat hump’ as described by Morse (1979, p. 579; 1986, figs 4 and 5). The roof interface is the locus of another, local, thermal maximum, but it is adiabatically cooler than the floor interface. Both interfaces, however, are emitters of latent heat. The floor interface cools to a minor degree out through the floor, and to a major degree out through the mushy zone. The roof interface cools through the latent heat-emitting, high-porosity, mushy zone at the roof and up through the country rock to the Earth's surface.
|
MUSH THICKNESS |
|---|
McBirney et al. call attention to the large temperature difference used in our analysis of mush thickness in Holness et al. (2007b
), and rightly complain that this would imply an unacceptably large gradient of floor cooling. The large thermal gradient between the crystal mush and subsolidus cumulate arose from our tacit assumption that a single uniform closure temperature might apply over a wide range of stratigraphy; this notion clearly cannot be correct, particularly in light of Fig. 1b and our earlier comment that the driving force for maturation is a function of the magnitude of the departure from equilibrium. On reflection, we think that the difference in temperature between the start of maturation and closure is special to each cumulus arrival. The observations of Holness et al. (2005) demonstrate that, for the special case of a Rum-sized magma body and junctions between clinopyroxene and olivine, this closure temperature is extremely high, essentially indistinguishable from the temperature at which clinopyroxene oikocrysts cease growing and therefore very close that of the arrival of clinopyroxene on the liquidus. The appropriate closure temperature for the Aug+ step in the Skaergaard Layered Series may likewise be very close to the trapping temperature. Each step-change will have its own, near-solidus, closure temperature.
Even without a quantitative estimate of the closure temperature, we have the firm observational evidence that less mature dihedral angles on intercumulus augites underlie more mature ones on cumulus augites that occur only a meter or so above at Aug+. This observation is indeed compatible with the field evidence for a thin mushy zone so meticulously demonstrated by Irvine et al. (1998).
|
HYPOTHESIS TESTING |
|---|
McBirney et al. conclude that to develop a new technique it must be tested against ‘examples of known origin’; in cumulate rocks this perception appears to be subject to some degree of difference on the part of the beholder. It was our impression that the cumulates of both Skaergaard and Rum were in some degree ‘of known origin’ and that our measurements could and did provide some elaboration on our knowledge of them. The combined result of our observations and their interpretation generates, in our opinion, a rich new field for quantifying the processes of solidification in cumulate rocks. We believe in the power of multiple working hypotheses and the importance of their falsification when that can be done. Falsification requires more than criticism, and to be persuasive, a new hypothesis has to be shown to be superior to the ruling hypothesis. The comments of McBirney et al. fall far short of this.
|
CONCLUSIONS: HOW THE LIQUIDUS EVENTS AFFECT THE MATURATION OF TEXTURES—A SUMMARY |
|---|
The field observations cited above for cyclic units in Rum and the basal section of Skaergaard firmly link changes in dihedral angle to observed heating events in the form of magma recharge. For the Skaergaard Intrusion above HZ, the following series of statements comprise the logic path of the enthalpy effects consequent to the arrival of a new liquidus phase.
- There is a durable, step-wise, increase in fractional latent heat released upon the arrival of each new phase at the liquidus.
- The step-change in fractional latent heat is associated with a step-change in the slope of the liquidus and a consequent increase of the crystal productivity as a function of temperature, as defined on p. 2369 in Holness et al. (2007b).
- This new higher fractional latent heat makes a greater contribution to the heat being transferred up through the mushy zone, out into the convecting liquid, and thence out through the roof.
- This greater contribution of latent heat to the enthalpy budget means that the subsolidus region below the solidification front is kept hotter for a longer time than it would have been without the arrival of a new phase. We reiterate the fundamental constraint that the total enthalpy loss from the system is nominally constant for small time intervals and that the local effect of temperature vs time is responsive to the ratio of latent to sensible heat.
- This duration of excess warmth allows the further maturation of the cumulate beyond that in the earlier cumulates before the addition of the new phase.
|
ACKNOWLEDGEMENTS |
|---|
We are most grateful to Mark Ghiorso for taking the time to guide us through the calculation of the fractional latent heat shown in Fig. 1, and for his enlightened remark, quoted above. We also thank John Brady for assistance with this calculation.
*Corresponding author. E-mail: marian{at}esc.cam.ac.uk
|
REFERENCES |
|---|
Ghiorso MS. Thermodynamic models of igneous processes. Annual Review of Earth and Planetary Sciences (1997) 25:221–241. doi: 10.1146/annurev.earth.25.1.221.[CrossRef][Web of Science]
Hess HH. Extreme fractional crystallization of basaltic magma: The Stillwater igneous complex. Transactions of the American Geophysical Union (1939) 3:430–432.
Holness MB. Spatial constraints on magma chamber replenishment events from textural observations of cumulates: the Rum Layered Intrusion, Scotland. Journal of Petrology (2005) 46:1585–1601.
Holness MB. Melt–solid dihedral angles of common minerals in natural rocks. Journal of Petrology (2006) 47:791–800.
Holness MB. Textural immaturity of cumulates as an indicator of magma chamber processes: infiltration and crystal accumulation in the Rum Layered Suite. Journal of the Geological Society, London (2007) 164:529–539.
Holness MB, Cheadle MJ, McKenzie D. On the use of changes in dihedral angle to decode late-stage textural evolution in cumulates. Journal of Petrology (2005) 46:1565–1583.
Holness MB, Nielsen TFD, Tegner C. Textural maturity of cumulates: a record of chamber filling, liquidus assemblage, cooling rate and large-scale convection in mafic layered intrusions. Journal of Petrology (2007a) 48:141–157.
Holness MB, Tegner C, Nielsen TFD, Stripp G, Morse SA. A textural record of solidification and cooling in the Skaergaard Intrusion, East Greenland. Journal of Petrology (2007b) 48:2359–2377.
Holness MB, Anderson AT, Martin VM, Maclennan J, Passmore E, Schwindinger K. Textures in partially solidified crystalline nodules: a window into the pore structure of slowly cooled mafic intrusions. Journal of Petrology (2007c) 48:1243–1264.
Holness MB, Hallworth MA, Woods A, Sides RE. Partial melting and metasomatism of cumulates by intrusive magma replenishment: the Wavy Horizon, Isle of Rum, Scotland. Journal of Petrology (2007d) 48:563–587.
Irvine TN. Heat transfer during the solidification of layered intrusions. I. Sheets and sills. Canadian Journal of Earth Sciences (1970) 7:1031–1061.
Irvine TN, Andersen J. C. Ø., Brooks CK. Included blocks (and blocks within blocks) in the Skaergaard intrusion: geological relations and the origins of rhythmic modally graded layers. Geological Society of America Bulletin (1998) 110:1398–1447.
Jang YD, Naslund HR. Major and trace element composition of the Skaergaard plagioclase; geochemical evidence for changes in magma dynamics during the final stage of crystallisation of the Skaergaard intrusion. Contributions to Mineralogy and Petrology (2001) 140:441–457.[CrossRef][Web of Science]
Maaløe S. Zoned plagioclase of the Skaergaard Intrusion, East Greenland. Journal of Petrology (1976) 17:398–419.
Morse SA. Kiglapait geochemistry I: Systematics, sampling, and density. Journal of Petrology (1979) 20:555–590.
Morse SA. Adcumulus growth of anorthosite at the base of the lunar crust. Journal of Geophysical Research (1982) 87:A10–A18.[CrossRef]
Morse SA. Convection in aid of adcumulus growth. Journal of Petrology (1986) 27:1183–1215.
Morse SA. Basalts and Phase Diagrams. (1994) Malabar, FL: Krieger.
Renner R, Palacz Z. Basaltic replenishment of the Rhum magma chamber: evidence from unit 14. Journal of the Geological Society, London (1987) 144:961–970.
Tait SR. Fluid dynamic and geochemical evolution of cyclic Unit 10, Rhum, Eastern Layered Series. Geological Magazine (1985) 122:469–484.[Abstract]
Toplis MJ, Brown WL, Pupier E. Plagioclase in the Skaergaard intrusion. Part 1: Core and rim compositions in the layered series. Contributions to Mineralogy and Petrology (2008) 155:329–340.[CrossRef][Web of Science]
Wager LR, Brown GM, Wadsworth WJ. Types of igneous cumulates. Journal of Petrology (1960) 1:73–85.
Wyllie PJ. Effects of the changes in slope occurring on liquidus and solidus paths in the system diopside–anorthite–albite. Mineralogical Society of America, Special Paper (1963) 1:204–212.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Tegner, P. Thy, M. B. Holness, J. K. Jakobsen, and C. E. Lesher Differentiation and Compaction in the Skaergaard Intrusion J. Petrology, May 1, 2009; 50(5): 813 - 840. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||