Hi Jordan,
Thanks for engaging. You raise some good questions.
First, I’m not sure whether the majority of philosophers of chemistry hold to the reductionist position - the SEP article gives representative arguments for both sides. There are also differing degrees of reductionism - strong vs. weak and so on.
In any case, if one holds to a non-reductionist position (Thomistic or not), then that means that chemistry has a set of unique laws which are irreducible to that of the laws of physics. This does not necessarily mean that these laws violate the lower-level laws of physics, only that they are not derivable from the latter. The higher-level laws are “just there”; by definition, they cannot be explained in terms of other laws.
More specifically, Aristotelian philosophy of science is at its core opposed to the philosophy of atomism [1]. In the conventional reductionist picture, the lowest level “building blocks” of matter - atoms (or elementary particles, or strings) are the most fundamental and the entities that actually exist. Higher-level entities such as bodies of water and chairs and humans do not really exist as a single “thing” (or substance), but are only an aggregate of these atoms. Chemical laws are only aggregates of the laws of physics.
In contrast, Aristotelian philosophy reverses this bottom-up picture to take a top-down picture. Here, macro-level objects [2] exist actually, while their constituents do not “really” exist; they exist only virtually. This doesn’t mean that the Aristotelian denies the utility of modern scientific theories about the atom. Rather, she tends to take a non-realist approach to the reality of entities such as atoms and molecules. By non-realist, I do not mean instrumentalist; Feser for example defends a structural realist view of philosophy of science, where scientific laws capture something real about the structure of reality but not every entity in these laws actually exist.
Here, I’m only summarizing some of the points made by Bishop in his paper Whence Chemistry?. I strongly advise you to read that paper in entirety if you want to get a fuller picture of the non-reductionist (not necessarily Aristotelian) position. In particular, Bishop’s point is that the Hamiltonian for different isomers are identical, since they only contain different terms corresponding to nucleons and electrons (and interactions between them) added together, without mathematics to indicate the molecular structure:
No, Aristotelians do not think that atoms are immutable. But that’s precisely the point of Aristotelianism - a hydrogen atom can undergo smaller-scale changes while remaining a hydrogen atom. The basic metaphysical philosophy of Aristotle is that it’s a middle ground between the two extremes of Heracliteanism (where everything changes, there is nothing constant) and Parmenideanism (where nothing changes, and everything is constant). In Aristotelianism, sometimes things can change while remaining the same (this is called an accidental change), while at other times things do change so much to the point that they are a totally different thing (substantial change).
In contrast, philosophical atomists do tend to fall into the Parmenidean extreme, where (philosophical) atoms are thought of as immutable. If you’re a hardcore reductionist, then you would probably believe something along the lines that (chemical) atoms are really just aggregates of smaller, immutable fundamental particles (i.e. quarks and electrons, or even strings) which are rearranging themselves according to some immutable laws of nature.
Going back to the question of H2O. As you asked, does a hydrogen atom cease to be a hydrogen atom when it becomes part of H2O? Now, it just happens to be that most Aristotelian-Thomists tend to think that the answer is yes. But this is by no means logically entailed by Aristotelian philosophy. The answer depends, of course, on what it means for a hydrogen atom to be a hydrogen atom. Or, in A-T terms: when hydrogen reacts with oxygen to form H2O, does it undergo a substantial change? To answer this question we cannot simply argue from the armchair, but also take into account actual empirical data. Currently our data says that there seems to be properties of H2O which are not possessed by hydrogen at all when it is existing by itself. This seems to point to the answer that yes, hydrogen does undergo a substantial change when becoming H2O.
Now, if someone, whether through empirical investigation or argument, found out that the properties of water are indeed fully reducible to that of hydrogen and oxygen, then the Aristotelian would answer that yes, H2O is not a unified entity (or substance) by itself, but only an aggregate of hydrogen and oxygen, similar to two rocks being piled on top of each other. Note, however, that this wouldn’t disprove A-T philosophy, only change the A-T interpretation of H2O. It’s important to understand that A-T philosophy is not being offered here as a competitor to scientific method and facts and, but as a competitor to meta-interpretations of scientific facts.
At least on face value, it doesn’t. You should read the Koons paper to understand his argument. My very crude recap of his argument: in statistical mechanics, we commonly start by summing over a discrete number of states. At some point, we change our summation signs to an integral, taking the limit of an infinite number of states. Most of actual statistical mechanics is done by integrating, not discrete summing. Koons argues that when this switch (from sums to integrals) happens, a more significant change occurs - the transition from the micro to the macro-level, which has behavior irreducible to the micro-level. He advances arguments that you cannot obtain this continuum limit by simply adding discrete states, invoking the Stone–von Neumann theorem in the process.
Endnotes
[1] Note that I am referring to atomism as a general philosophy of science, not necessarily the modern scientific model of the atom that formed in the early 20th century with Rutherford, Bohr, etc.
[2] The majority of Thomists seem to only things as large and tangible as human beings and chairs exist, or molecules can exist. However, I don’t see in principle any obstacle against taking a hybrid view, where say, molecules exist actually but quarks and electrons (mostly) exist only virtually.