Wednesday, August 03, 2005

The Manifest Image of Chemistry (LFPA)

Citation: J. van Brakel, "Chemistry as the Science of the Transformation of Substances," Synthese (1997) 111:253-282.

Summary: In this article, van Brakel develops an argument against the reduction of chemistry and thermodynamics to microphysics; he does this by focusing on the macroscopic level of substances and their transformations and providing "some preliminaries for assessing whether such macroscopic notions as chemical substance, equilibrium, and temperature can be reduced to microphysics, here defined to include statistical physics, atomic physics, particle physics, and quantum mechanics" (254). This argument is part of a larger argument about whether (in Sellars's terms) the scientific image can replace the manifest image; but the article stays within somewhta narrower bounds than this.

Historical Background. Van Brakel begins by suggesting that the common view that chemical thought about chemical compounds (pure substances) derives in a more-or-less straight line from ancient atomism via Boyle is false. Rather, he suggests, the real source of the notion of pure substance derives from the work of early modern metallurgists and apothecaries. Corpuscularianism is actually inconsistent with chemical understandings of element and compound; it "only covers compounds of corpuscles, without taking into account the relation of the notion of chemical compound to chemical synthesis and analysis, the combination and separation of substances, and concepts like conservation, reversibility, and homgeneity" (255). If this is true, physical and chemical atomism need to be regarded as distinct; the latter is due to Dalton, not Boyle. The reason why the two are so often simply identified is very likely that people assume that that the ultimate ontology is what physics says it is; but, as van Brakel points out, the issue is more complex -- even if physically speaking there are no chemical atoms, it wouldn't follow that chemistry is non-referential.

Ordinary and Scientific Water. Consider Hare's argument that ordinary liquid water does not supervene on H2O because (1) 'water' and 'H2O' are words with different senses; and (2) the notion of natural kind or substance that underlies such alleged supervenience is a recent discovery. Hare's support for (1) is that if people were thirsty and asked for water, they wouldn't be pleased if you directed a jet of steam at them; but van Brakel points out that they wouldn't be pleased if you directed a stream of liquid water at 98 degrees Celsius, or a stream of sea-water. Van Brakel then goes on to point out that the commonsense notion of water, and the notion of water as a pure substance, have a very long history (Aristotle argues for a water cycle, for instance, which requires that water be a substance capable of undergoing phase transformation).

Although it is correct that not all uses of "water" imply that it is H2O, all uses of "water" do imply that it is a natural kind of the pure substance type. Water in all its modifications (liquid, solid, vapour) is the same substance. This supports the point that knowledge about "materials and their transformations" is more robust than the local microphysical picture of the moment. (258)


So when we say that something is water we are saying something about its essence, although this essence needn't be a microscopic structure. Concepts like water and substance are better entrenched than concepts like atom and molecule. The manifest image is prior to the scientific image; scientific descriptions like H2O refer to what is ordinarily described as water, not the other way around.

Reduction of Physico-Chemical Thermodynamics to Microphysics? An interesting question to ask is whether temperature is reducible. The macroscopic concept of temperature presupposes the notion of equilibrium, which operationally presupposes no significant changes in macroscopic features. What happens when one goes microscopic, then? At that level there is no such thing as equilibrium, so we can't define temperature in terms of the 'zeroth law of thermodynamics'. Even strong reductionists like Churchland are occasionally puzzled about how temperature would reduce. The first temptation would be simply to identify temperature with mean kinetic molecular energy, but there are any number of problems with that: such an interpretation works out well for ideal gases, but not for solids, plasmas, or a vacuum; quantum mechanics implies that for a dengse gase at low temperature kinetic energy is related not only to temperature but to density as well, and it is not even related to temperature by simple proportionality; and even if temperature could be reduced, it doesn't follow that concepts like boiling point or other concepts related to phase transformation could be. There are no molecular definitions of phases.

The ultimate basis for this sort of problem appears to be that there is at present no viable way to reduce the immensely important concept of equilibrium, in part because of the time asymmetry. Entropy in thermodynamics is both additive and non-decreasing; Boltzmann's entropy is only additive; Gibbs's entropy is only non-decreasing. In a hundred years no reduction of it has been possible. Nor is this all:

As a last example consider surfaces. For surfaces the macroscopic, thermodynamic surfaces ar emuch more real than surfaces at the sub-macroscopic level. IF we don't look at the molecular/atomic level, we can ascribe energy to a surface and have forces work along surfaces and explain all kinds of things: for example, why drops from a tap have a certain size....On the other hand, at the molecular level it's not at all clear what the surface is. (262)


This same general sort of problem, that the reduction is impossible without smuggling in the macrophysical concepts, appears elsewhere: not only in temperature, but in colour and schizophrenia, as well as to molecular chemistry, to which van Brakel now turns.

Reduction of Chemistry to Physics via Quantum Chemistry? The notion of structure has different meanings at different levels of description; while quantum mechanical structurs have been used to make excellent predictions of chemical properties, this isn't sufficient to prove that complete reduction is possible in principle, because (again) the higher-level properties get smuggled into the alleged reducing theory. Thus, for instance, quantum chemistry is to a great measure based on the Born-Oppenheimer approximation, which itself is grounded in the classical chemical view, "the picture of a semi-rigid framework of atoms connected by bonds that rotate and translate in ordinary space as time elapses" (265). Likewise, you cannot get the concept of electronic configurations from quantum mechanics; you only get them from spectral observations. And so forth. As van Brakel points out, "numerical methods are governed by what experimental data have to be predicted; or, if new predictions are made, the choice of parameters is governed by extrapolations made from calculations performed on other molecules or general experience derived from experimental work on chemical compounds and reaction mechanisms" (266). Quantum chemistry assumes the existence of a molecule, and adapts quantum mechanics to that; it does not pull the molecule out of quantum mechanics, and, in particular, (as is commonly noted in philosophy of chemistry) there is no clear way you can get chemical structure (molecular shape) out of quantum structure. The view that chemistry is simply reducible to physics is based on the assumption that science is heading toward microreductionism; it is not based on any actual success in the reduction.

Essences? It is common to assume that 'the concept/laws of water can be reduced to the concept/laws of H2O' is equivalent to saying 'the essence of water is H2O'. However, this is not so clear. One could say that liquid water and water vapor are the same substance because they are both H2O, but (as already pointed out) we could simply say that they are the same substance because if water is evaporated and then condensed again, the water comes back. In other words, one can identify sameness of substance without appeal to microphysical structure, just as one can identify butterflies and caterpillars as different forms of the same species without appealing to genetics. The common response to this is to say that the microphysical structure doesn't merely identify sameness of substance, but explains it. The problem is that it isn't clear that it can actually do so. Take a strong case. What would it mean to say that two substances are molecularly identical? Do we include the velocities and relative positions of molecules? If so, there are no molecularly identical substances; velocities and relative positions of molecules are in a state of continual change. Do we leave them out? Then we have to ignore issues related to things like temperature.

Further, it is simply false to say that the essence of water is H2O. In liquid water there are H3O+ and OH- ions, which are absent from water vapor; in water vapor there are H4O2 and other H2O polymers that are not always found in liquid water. The microstructure of water actually depends on the context. And water is a relatively simple case. Many chemical properties cannot be understood in terms of molecules (acidity, conductivity, etc.); large molecules are only approximately the same; etc. The upshot of it is, we can't define chemical natures in terms of constitutive structures unless we assume that the constituting atoms have or are essences; and there are immense problems with determining what an atom-essence would be. The numerical part of an atomic number is derived from the number of protons/electrons; but it is not identical to this (Krypton-36 is not identical to thirty-six hydrogen atoms). The electrons, etc., must be configured and related to each other in a certain way, and what determines the significant relations is our having already identified the substance.

(In a footnote, van Brakel makes an important distinction between unification and reduction. One can hold that chemistry and physics are unified without holding that one is reduced to the other.)

Evaluation: Van Brakel makes a number of points (not all of which I have done complete justice to), some of which I'm not particularly competent to evaluate. But there are a number of things that are dead-on. The point about structures is important, as is the distinction betwen unification and reduction. If someone, for instance, were to parse reductionism about a system in terms of parts and all their relations, this would be a double equivocation, since it would equivocate about the different sorts of relations parts can have at different levels, and it would equivocate about the reduction itself: you may have a unification, but you don't have a reduction unless the higher-order relations reduce to lower-order relations. And the point about H2O not being the essence of water is also important (since the issue is not just water but any substance). When I decided I was going to do another LAFP, I chose this article by van Brakel because it shows something that I've held for a long time, namely, that there's great work being done in the philosophy of chemistry that shouldn't be ignored by those interested in philosophy of science (which tends to revolve around the two poles of physics and biology, dismissing chemistry completely).