EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Leveraging crystal-scale data to constrain the conduit flow regime in persistently active volcanoes

Jenny Suckale, Michelle diBenedetto, and Zhipeng Qin
Jenny Suckale et al.
  • Standford University, United States

Persistently active volcanoes are often closely monitored, yielding a rich archive of observational data. The availability of varied observations provides a unique opportunity for improving theoretical models of magma dynamics, but data and model can be difficult to compare directly. Geophysical observations like seismicity or geodetic measurements often operate at similarly large scales as many models, but they only provide indirect and non-unique testimony of the processes occurring at depth. In contrast, crystals in erupted tephra or scoria samples record at least some aspects of the pre-eruptive condition in the volcanic conduit directly, but refer to spatial scales that are much smaller than most models resolve.

The goal of this paper is to demonstrate the potential of crystalline-scale data for distinguishing directly between different conduit-flow models. As a proof of concept, we focus on the preferential alignment of olivines crystals from tephra erupted at Kilauea Iki in 1959. Prior petrographic analysis suggests that these olivine glomerocrysts formed through synneusis of individual crystals. To evaluate the fluid-dynamical conditions under which both crystal synneusis and preferential crystal alignment would occur, we compare two broad classes of conduit flow models, unidirectional and bidirectional models.

We hypothesize that the observed preferential alignment of olivine crystals is created by a pronounced, nearly stationary wave at the interface that separates the ascending and descending magmas in bidirectional flow models. Crystals in bidirectional flow are hence exposed to a superposition of wave and shear, while crystals in a unidirectional, laminar flow experience approximately constant shear strain during ascent. To test our hypothesis, we quantify the crystal alignment resulting from a pure shear flow and from the superposition of a stationary wave on shear flow through two complementary model approaches. We first derive an analytical model for when crystals align under the joint influence of a wave and shear flow. We then use direct numerical simulations to quantify how crystal-crystal interactions modulate the analytically predicted preferential alignment of crystals.

We find that the formation of glomerocrysts with preferential aligned olivine crystals is consistent with bidirectional flow models, but unlikely to form in a unidirectional model. We emphasize that the imprint of the conduit flow on the crystals is subtle, suggesting that both clustering or alignment in isolation would be compatible with a much wider range of flow conditions than the observed conjunction of both attributes in the Kilauea Iki olivines. To our knowledge, these observations provide the first direct evidence of bidirectional flow in volcanic conduits.

How to cite: Suckale, J., diBenedetto, M., and Qin, Z.: Leveraging crystal-scale data to constrain the conduit flow regime in persistently active volcanoes, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-22411,, 2020

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 1 – uploaded on 05 May 2020
  • CC1: Comment on EGU2020-22411, Thomas Griffiths, 05 May 2020

    Dear Jenny, this work is really fantastic! I have never worked on these rocks directly but I study crystal clustering, especially orientation relationships. Ever since I read the Schwindinger 1999 paper and now again with the new paper by Wieser et al 2019 I have been wondering if fluid dynamics (forgive me if it's not the right term) can explain the full orientation relationship or not. I have a lot of questions, I hope not too many...

    The most interesting thing for me is your work contradicts one idea I had about the formation of the oriented clusters, which is that by tumbling and trying many possible orientation relations the crystals eventually randomly find one which "sticks", perhaps due to surface energy or growth rates. Instead, it seems that crystals might only have one or a few chances, making it more surprising to me that the orientation relatinships are so close.

    My first question: As far as I understand, the Kilauea olivine clusters are not just aligned facet to facet but at specific misorientations around those paired facets. Do you think this full alignment could be explained by just fluid interactions, perhaps if you had a 3d model? Or do we still need something else to help out finding the perfect relation?

    And my second (maybe easier) question: Do we know if the clusters we see represent a low or a high proportion of the total number of olivine phenocrysts in this zone of the conduit? Either from your modelling or the frequency of non-clustered olivines seen alongside clustered ones in the Kilauea scoria. I wonder if we are focusing on a low number of special cases, or if a significant fraction of olivines aggregate this way?

    best regards,

    Tom Griffiths, University of Vienna (Austria)


    • AC1: Reply to CC1, Jenny Suckale, 05 May 2020

      Dear Tom,

      thanks for your interest in this work. It looks like we are interested in similar things, but coming from different angles. The very short answer to your question is that I don't think that fluid dyanics can explain the orientational relationship fully. Crystal shape is really consequential and surface energies etc matter, too. Here, we are looking "only" at the fluid dynamical component, not because it's the only thing that matter, but to test how much of the data we can explain with that process alone.

      To your first question: I agree that the specific crystal geometry plays an important role and yes, alignment in the Kilauea Iki samples occurs along in a facet-to-facet way (as shown nicely by Wieser et al., 2019). That is probably the primary reason why we don't actually reproduce the data exactly. We simplify the actual shape of the olivine crystals to ellipsoids or rectangles to understand how much of the alignment behavior is not due to the specifics of the crystalline facets. After all, there are plenty olivine crystals in other flows that do not show this kind of clustering behavior, so there seems to be something special about the flow field sampled by the Kilauea Iki olivines.

      To the second question: Schwindinger says in her thesis (and later in the 1989 paper) that a bit more than half of the olivines are clustered. Many of the non-clustered ones are relatively small. If we assume that the crystals would never stick, we cannot reproduce that high percentage (at the low crystal fraction where we see the large-misorientation angles, the percentage of clustered crystals is only in the 20% range). That being said, if we assume that crystals would eventually intergrow after being in contact for an extended period of time, you can easily get to a higher percentage of clustered crystals. 

      Regarding your idea of "sticking": I don't think sticking would be instantaneous in the sense that a brief collision could lead to sticking, because crystals do not actually tend to touch each other when colliding. The reason is that the viscosity of the surrounding magmatic fluid is pretty high and it takes a lot of force to fully displace this fluid. Instead, I think the crystals grow together over time by both encroaching on the thin film that originally separated them. Once the crystals are stably clustered, there is no longer any motion in the thin fluid film separating these crystals (the flow goes around the cluster rather than through) and diffusion becomes the dominant process which helps the intergrowth. That's also where the crystalline facets become important.

      Does that make sense? 


      • CC3: Reply to AC1, Thomas Griffiths, 06 May 2020

        Dear Jenny, thanks very much! So it seems like this process is perhaps relatively efficient at building clusters, given enough time, unless there is some further sampling bias we are missing. Which throws up another question - do we have any idea of the timescale of the cluster formation event? If its in a conduit I guess it can't be that long, geologically?

        I think I am still not quite following one thing though - you refer to the crystal shape as not being part of the fluid dynamic reasons for clustering. Maybe there is a terminology misunderstanding for me. I thought that the shape does impact the fluid dynamics, perhaps providing the extra tiny level of complexity leading to the emergence of perfect alignment.

        But in any case what I am taking from your answer is that even though the shape is a key factor, there doesn't necessarily need to be a long or short range force due to surface energies that is the final thing that gets us from a close alignment to a very specific orientation relationship. The particular flow conditions bring crystals into sustained contact, and the euhedral shapes mean that they assume very fixed orientation relationships, then over time the two crystals get joined.

        Does any of that seem right?

        best regards,

        Tom Griffiths

        • AC3: Reply to CC3, Jenny Suckale, 06 May 2020

          Dear Tom,

          yes that sounds about right. Let me clarify the role of shape: There is no doubt that shape is important, but shape entails various things. For example, there is the aspect ratio and then there is the detailed crystallographic facies that an axisymmetric crystal might have. We show in our model that the axisymmetry is the key factor governing how crystals align in a wave. Interestingly, this is not "just" a question of axisymmety or not, we can compute the preferential alignment angle for each specific aspect ratio, e.g., crystals with an aspect ratio of 2:3 would prefer a different angle than crystals with aspect ratio of 2:4. Our model shows that crystals with the same aspect ratio assume a very similar angle if one is elliptic and one is rectangular. So, for the alignment angle, the crystallographic facies are not very important, but for the misorientation angle they are (alignment and misorientation angles are closely related but not the same: the alignment angle is a consequence of the fluid-crystal interactions, the misorientation angle is a consequence primarily of crystal-crystal interactions). Crystal faces are important for explaining the clustering of misorientation angles. E.g., Wieser et al. find that a misorientation angle of 80.9 is very common in the Kilauea Iki olivines. Our analysis would suggest that the preferential alignment angle for the aspect ratio of the Kilauea Iki olivines is roughly in the 34-40 degree range (it obviously depends on the wave conditions too, so can't give a precise value without knowing more about the wave), which means that many crystals in the flow will be separated by an angle of 68-80 (because axisymmetry implies a set of two favorable angles). Not a bad fit, but a bit on the low end and I think that is because we do not consider the specific crystallographic facies. For the sake of argument: Let's assume that the flow will tend to align the crystals at say 79 degree (with some wobble around that angle). That is very close to the 80.9 angle that separates  two 100 facies, so these two crystallographic facies will be in contact for many crystal pairs and hence overgrow with time. So, the overgrowth is facilitated by the similarity between the preferential alignment angles and the angle between the two 100 facies (and in some sense helped by the wobble around the preferential alignment angle). I think that the time scale of overgrowth depends primarily on the diffusivity of the different chemical components I need for crystal growth in a similar way as a crystal growing in a stagnant liquid. So yes, definitely short on geological time scales. Does that make sense?


          • CC4: Reply to AC3, Thomas Griffiths, 07 May 2020

            Yes, that sounds good, I think this part especially emphasises the links well for me:

            "Let's assume that the flow will tend to align the crystals at say 79 degree (with some wobble around that angle). That is very close to the 80.9 angle that separates  two 100 facies, so these two crystallographic facies will be in contact for many crystal pairs and hence overgrow with time."

            Fluid dynamics gets the crystals almost all the way there and crucially also holds them in place, so that the physical shape and just possibly a tiny bit of surface enrergy can reach the final misorientation.

            One last thing that I am missing: Is it just trivially obvious that the long axes of the olivines align, due to the way the tabular crystals orient in the flow? There are many ways to stick the same two facets together, with full rotational freedom.

            • AC4: Reply to CC4, Jenny Suckale, 07 May 2020

              Right! This "Fluid dynamics gets the crystals almost all the way there and crucially also holds them in place, so that the physical shape and just possibly a tiny bit of surface energy can reach the final misorientation" is a good way of thinking about it, in my opinion.

              Regarding your question about the alignment along the long axis: When axisymmetric crystals align along their long axes, they form a sandwich-type of arrangement, which minimizes surface area of the cluster and hence tends to reduce hydrodynamic drag exerted on the crystals by the flow. In a unidirectional flow, this is the arrangement that most closely follows the stream lines of the flow (eg think of a windtunnel experiment  - a crystal sticking straight up would experience significant force from the flow and topple over). Since at the scale of small crystals, many flows look approximately unidirectional, this type of arrangement is common. In a wave-dominated flow, however, the streamlines look very different (you now have vorticity), so the crystals arrange differently. The bottom line though is that you can get the full range of alignment angles. In our paper, we derive a non-dimensional number that represents the ratio of mean shear to wave-induced shear over the aspect ratio of the crystal. Based on that, you can compute the preferential alignment angle for different flow fields and aspect ratios. That is also why the fact that we observe large misorientation angles in the Kilauea Iki olivines is such a valuable constraint: for the crystal geometry we see there, the only way you can get to the observed angles is with a pronounced wave in the background. Is that helpful?


              • CC5: Reply to AC4, Thomas Griffiths, 11 May 2020

                Dear Jenny,

                Yes, thanks, I think I have understood everything now, at least as much as one can in a non face to face discussion!

                Its a fascinating and for me a little surprising result, I always imagined (with no justification) some kind of magma chamber having to be involved, not a conduit.

                Along with several others, I am more curious than ever about the examples of Quarz phenocryst synneusis in high silica plutonic rocks and how these may occur.

                Thanks again for explaining everything.

                best regards,

                Tom Griffiths

  • CC2: Comment on EGU2020-22411, Penny Wieser, 05 May 2020

    This is really interesting Jenny! It makes a lot more sense for crystal clusters to be forming this way than just during passive settling in chambers. I wonder if the reason  we see more "peaked" distributions in natural samples vs. your models (, Fig. 3) is that olivine only has a set number of faces, so if the alignment imposed by the flow doesn't equal one of those "magic" misorientation angles where the faces have a high degree of contact, crystals don't have a chance to cement together by growth, and they continue to move on and rotate until one of those angles is met? Out of interest, does your model recreate our observations that almost all clusters are misorientated about [100]? This was quite surprising to us at the time, and it would be nice to have an explanation for it!

    • AC2: Reply to CC2, Jenny Suckale, 05 May 2020

      Yes, I agree that the more peaked distributions you see in natural samples is due to the specific crystalline faces. There is an important interplay between flow and crystal shapes though. The preferential alignment angles depend not only on the flow, but also on the crystal shape, primarily the aspect ratio. I would argue that if the flow brings the crystals close to an angle at which their faces can align, intergrowth can (and likely will) happen. As you might remember from the presentation, the crystals actually wobble around the preferential alignment angle, so you don't have to hit the magic angle exactly (this wobble tends to be quite sensitive to the crystalline faces). Regarding your question regarding preferential clustering on 100: Yes, that's exactly what we are trying to explain and what (in my opinion) our analytical model explains quite beautifully. I would argue that the large misorientation angle represents the set of two angles that represent the preferential alignment of crystals in a flow field with shear and a steady wave superimposed.