CHAPTER THREE

A FIRST LOOK AT
THERMODYNAMIC EQUILIBRIUM

OVERVIEW our notion of temperature is associated with this ordered

arrangement, so that we speak of items having higher or
In this chapter, we introduce the foundations of thermo- lower temperatures than other items, depending on how
dynamics. A major goal is to establish what is meant by “hot” or “cool” they feel to our senses. In the more rig-
the concept of equilibrium, described informally in the orous terms we must use as geochemists, though, this
context of the party example in chapter 1. The idea of casual definition of temperature is inadequate for two
equilibrium is an outgrowth of our understanding of reasons. First of all, a practical temperature scale should
three laws of nature, which describe the relationship have some mathematical basis, so that changes in tem-
between heat and work and identify a sense of direction perature can be related to other continuous changes in
for change in natural systems. These laws are the heart system properties. It is hard to use the concept of tem-
of the subject of thermodynamics, which we apply to a perature in a predictive way if we have to rely on disjoint,
variety of geochemical problems in later chapters. subjective observations. Second, and more important, it
To reach our goal, we need to examine some familiar is hard to separate this common perception of tempera-
concepts like temperature, heat, and work more closely. ture from the more elusive notion of “heat,” the quantity
It is also necessary to introduce new quantities such as that is transferred from one body to another to cause
entropy, enthalpy, and chemical potential. The search changes in temperature. It would be helpful to develop a
for a definition of equilibrium also reveals a set of four definition of temperature that does not depend on our
fundamental equations that describe potential changes in understanding of heat, but instead relies on more familiar
the energy of a system in terms of temperature, pressure, thermodynamic properties.
volume, entropy, and composition. Imagine two closed systems, each of which is homo-
geneous; that is, each system consists of a single substance
TEMPERATURE AND with continuous physical properties—from now on, we
EQUATIONS OF STATE will call this kind of substance a phase—that is not under-
going any chemical reactions. If this condition is met,
It is part of our everyday experience to arrange items the thermodynamic state of either system can be com-
on the basis of how hot they are. In a colloquial sense, pletely described by defining the values of any two of its

35
36 G E O C H E M I S T R Y : PAT H WAY S A N D P R O C E S S E S

intensive properties. These might be pressure and viscos- equilibrium with C, then A is in equilibrium with C.”
ity, for example, or acoustic velocity and molar volume. Notice—this is important—that figure 3.1a could de-
To study the concept of temperature, let’s examine the re- scribe either system. We could just as well have varied
lationship between pressure and molar volume, using the pressure and molar volume in the other system and
symbols p and v̄ in one system, and P and V̄ in the other. found an infinite number of values of p and v̄ that lead
(Refer back to chapter 1 to confirm that molar volume, to thermal equilibrium with the system at P1,V̄1. These
the ratio of volume to the number of moles in the sys- values would define a curve on a graph of p versus v̄,
tem, is indeed an intensive property. Notice, also, that shown in figure 3.1b.
we are now beginning to use the convention of an over- Curves like these are called isothermals. What we
score to identify molar quantities, as described in chap- have just demonstrated is that states of a system have the
ter 1.) same temperature if they lie on an isothermal. Further-
If we place the two containers in contact, changes of more, two systems have the same temperature if their
pressure and molar volume will take place spontaneously states on their respective isothermals are in thermal equi-
in each system until, if we wait long enough, no further librium. These two statements, together, define what we
changes occur. When the two systems reach that point, mean by “temperature.” You could express these two
they are in thermal equilibrium. Let’s now measure their statements algebraically by saying that the each of iso-
pressures and molar volumes and label them P1, V̄1 and therms in our example can be described by a function:
p1, v̄1. There is no reason to expect that P1 will neces-
sarily be the same as p1 or that V̄1 will be equal to v̄1. In f(p, v̄) = t or F(P, V̄) = T,
general, the systems will not have identical properties. and that the systems are in thermal equilibrium if t = T.
If we separate the systems for a moment, we will find Functions like these, which define the interrelationships
it possible to change one of them so that it is described among intensive properties of a system, are called equa-
by new values P2 and V̄2 that still lead to thermal equi- tions or functions of state.
librium with the system at p1 and v̄1. There are, in fact, Because many real systems behave in roughly similar
an infinite number of possible combinations of P and V̄ ways, several standard forms of the equation of state
that satisfy this condition. These combinations define a have been developed. We find, for example, that many
curve on a graph of P versus V̄ (fig. 3.1a). Each combi- gases at low pressure can be described adequately by the
nation of P and V̄ describes a state of the system in ther- ideal gas law:
mal equilibrium with p1 and v̄1. All points on the curve
are, therefore, also in equilibrium with each other. This PV̄ = RT,
observation has been called the zeroth law of thermo-
dynamics: “If A is in equilibrium with B and B is in in which R is the gas constant, equal to 1.987 cal mol−1
K−1. Other gases, particularly those containing many
atoms per molecule or those at high pressure, are char-
acterized more appropriately by expressions like the Van
der Waals equation:

(P + a /V̄ 2)(V̄ − b) = RT,

in which a and b are empirical constants. We apply these

and other equations of state in chapter 4.
Our notion of temperature, therefore, depends on
the conventions we choose to follow in writing equations
of state. A practical temperature scale can be devised
simply by choosing a well-studied reference system (a
FIG. 3.1. (a) All states of a system for which f(P, V) = T are said
thermometer), writing an arbitrary function that remains
to be in thermal equilibrium. A line connecting all such states is
called an isothermal. (b) If another system contains a set of constant (that is, generates isothermals) for various states
states lying on the isothermal f(p, v) = t, and if t = T, then the of the system in thermal equilibrium, and agreeing on a
two systems are in thermal equilibrium. convenient way to number selected isothermals.
 A First Look at Thermodynamic Equilibrium 37

WO R K same at all places in the gas and the gas is not expanding
in mechanical equilibrium with its surroundings.
In general terms, work is performed whenever an object
is moved by the application of a force. An infinitesimal THE FIRST LAW OF THERMODYNAMICS
amount of work dw, is therefore described by writing:
During the 1840s, a series of fundamental experiments
dw = F dx,
were performed in England by the chemist James Joule.
in which F is a generalized force and dx is an infinitesi- In each of them, a volume of water was placed in an in-
mal displacement. By convention, we define dw so that sulated container, and work was performed on it from
it is positive if work is performed by a system on its the outside. Some of these experiments are illustrated in
surroundings, and negative if the environment performs figure 3.2. The paddle wheel, iron blocks, and other
work on the system. We know from experience that an mechanical devices are considered to be parts of the
object may be influenced simultaneously by several insulated container. The temperature of the water was
forces, however, so it is more useful to write this equa- monitored during the experiments, and Joule reported
tion as: the surprising result that a specific amount of work per-
formed on an insulated system, by any process, always
dw = ΣF dx .
i i results in the same change of temperature in the system.
The forces may be hydrostatic pressure, P, or may be di-
rected pressure, surface tension, or electrical or magnetic
potential gradients. All forms of work, though, are equiv-
alent, so the total work performed on or by a system can
be calculated by including all possible force terms in the
equation for dw. The thermodynamic relationships that
build on the equation do not depend on the identity of
the forces involved.
This description of work is broader than it often needs
to be in practice. Work is performed in most geochemi-
cal systems when a volume change dV is generated by
application of a hydrostatic pressure:

dw = P dV.

This is the equation we most often encounter, and geo-

chemists commonly speak as if the only work that
matters is pressure-volume work. Usually, the errors
introduced by this simplification are small. It is always
wise, however, to examine each new problem to see
whether work due to other forces is significant. Later
chapters of this book discuss some conditions in which
it is necessary to consider other forces.
Note that we have defined work with a differential
equation. The total amount of work performed in a pro-
cess is the integral of that equation between the initial
and final states of the system. Because forces in geologic
environments rarely remain constant as a system evolves,
FIG. 3.2. English chemist James Joule performed experiments to
the integral becomes extremely difficult to evaluate if we relate mechanical work to heat. (a) A paddlewheel is rotated and
do not specify that the process is a slow one. The work a measurable amount of work is performed on an enclosed water
performed by a gas expanding violently, as in a volcanic bath. (b) Two blocks of iron are rubbed together. (c) Electrical
eruption, is hard to estimate, because pressure is not the work is performed through an immersion heater.
38 G E O C H E M I S T R Y : PAT H WAY S A N D P R O C E S S E S

What changed inside the container, and how do we it tells us that the work done on a system by an adiabatic
explain Joule’s results? In physics, work is often intro- process is equal to the increase in its internal energy, a
duced in the context of a potential energy function. For function of the state of the system. We may conclude fur-
example, a block of lead may be lifted from the floor to ther that if a system is isolated, rather than simply closed
a tabletop by applying an appropriate force to it. When behind an adiabatic wall, no work can be performed on
this happens, we recognize not only that an amount of it from the outside, and its internal energy must remain
work has been expended on the block, but also that its constant.
potential energy has increased. Usually, problems of this Equation 3.1 can be expanded to say that for any
type assume that no change takes place in the internal system, a change in internal energy is equal to the sum
state of the block (that is, its temperature, pressure, and of the heat gained (dq) and the work performed on the
composition remain the same as the block is lifted), so system (−dw):
the change in potential energy is just a measure of the
change in the block’s position. When Joule performed dE = dq − dw. (3.2)
work on the insulated containers of water, however, it
Notice that dE is a small addition to the amount of in-
was not the position of the water that changed, but its
ternal energy already in the system. Because this is so,
internal state, as measured by a change in its tempera-
the total change in internal energy during a geochemical
ture. The energy function used to describe this situation
process is equal to the sum of all increments dE:
is called the internal energy (symbolized by E) of the
system, and Joule’s results can be expressed by writing dE = Efinal − Einitial = ∆E.

dE = −dw. (3.1) The value of ∆E, in other words, does not depend on
how the system evolves between its end states. In fact, if
Thus, there are two ways to change the internal en- the system were to evolve along a path that eventually
ergy of a system. First, as we implied when considering returns it to its initial state, there would be no net change
the meaning of temperature, energy can pass directly whatever in its internal energy. This is an important
through system walls if those walls are noninsulating. property of functions of state, described in more mathe-
Energy transferred by this first mode is called heat. If matical detail in appendix A.
Joule’s containers had been noninsulators, he could have In contrast, the values of dq and dw describe the
produced temperature changes without performing any amount of heat or work expended across system bound-
work, simply by lighting a fire under them. Second, as aries, rather than increments in the amount of heat or
Joule demonstrated, the system and its surroundings can work “already in the system.” When we talk about heat
perform work on each other. or work, in other words, the emphasis is on the process
To investigate these two modes of energy transfer, of energy transfer, not the state of the system. For this
we need to distinguish between two types of walls that reason, the integral of dq or dw depends on which path
can surround a closed system. What we have spoken of the system follows from its initial to its final state. Heat
loosely as an “insulated” wall is more properly called and work, therefore, are not functions of state.
an adiabatic wall. (The Greek roots of the word adia-
batic mean, appropriately, “not able to go through.”) Worked Problem 3.1
Adiabatic processes, like those in Joule’s experiments,
do not involve any transfer of heat between a system and Consider the system illustrated in figure 3.3, whose initial state
its surroundings. Perfect adiabatic processes are seldom is described by a pressure P1 and a temperature T1. This might
seen in the real world, but geochemists often simplify be, for example, a portion of the atmosphere. Recall that the
natural environments by assuming that they are adia- equation of state that we use to define isothermals for this sys-
tem tells us what the molar volume, V̄1, under these conditions
batic. A nonadiabatic wall, however, allows the passage
is. For ease of calculation, assume that the equation of state for
of heat. It is possible to perform work on a system that this system is the ideal gas law, V̄ = RT/P. Compare two ways
is bounded by either type of wall. in which the system might slowly evolve to a new state in which
Equation 3.1 is an extremely useful statement, re- P = P2 and T = T2. In the first process, let pressure increase
ferred to as the First Law of Thermodynamics. In words, slowly from P1 to P2 while the temperature remains constant, then
 A First Look at Thermodynamic Equilibrium 39

ENTROPY AND THE SECOND LAW

OF THERMODYNAMICS

Taken by itself, equation 3.2 tells us that heat and work

are equivalent means for changing the internal energy of
a system. In developing that expression, however, we
have engaged in a little sleight-of-hand that may have led
you, quite incorrectly, to another conclusion as well. By
stating that heat and work are equivalent modes of en-
ergy transfer, the First Law may have left you with the
impression that heat and work can be freely exchanged
FIG. 3.3. As a system’s pressure and temperature are adjusted for one another. This is not the case, as a few common-
from P1T1 to P2T2, the work performed depends on the path that the place examples will show.
system follows. A glass of ice water left on the kitchen counter grad-
ually gains heat from its surroundings, so that the water
warms up and the room temperature drops ever so
slightly. In the process, there has been an exchange of
heat from the warm room to the relatively cool water.
let temperature increase from T1 to T2 at constant pressure P2. As Joule showed, the same transfer could have been ac-
In the second process, let temperature increase first at constant complished by having the room perform work on the ice
pressure P1, and then let pressure increase from P1 to P2. (If this
water. It is clearly impossible, however, for a glass of
were, in fact, an atmospheric problem, the isobaric segments of
water at room temperature to cool spontaneously and
these two paths might correspond to rapid surface warming on
a sunny day, and the isothermal segments might reflect the pas- begin to freeze, although we can certainly remove heat
sage of a frontal system.) Is the amount of work done on these by transferring it first from the water to a refrigeration
two paths the same? system. In a similar way, a lava flow gradually solidifies
This problem is similar to problem 1.8 at the end of chapter 1. by transferring heat to the atmosphere, but this natural
If the only work performed on the system is due to pressure- process cannot reverse itself either. It is impossible to melt
volume changes, then the work along each of the paths is de-
rock by transferring heat to it directly from cold air.
fined by the integral of PdV̄. For path 1,
If we were shown films of either of these events, we
 
P2T1 P2T2
w1 = PdV̄ + PdV̄. would have no trouble recognizing whether they were
P1T1 P2T1
being run forwards or backwards. Equation 3.2, how-
Along path 2,
ever, does not provide us with a means of making this

 
P1T2 P2T2
w2 = PdV̄ + PdV̄. determination of direction theoretically. On the basis
P1T1 P1T2
of experience with natural processes, therefore, we are
To evaluate these four integrals, find an expression for dV̄ by driven to formulate a Second Law of Thermodynamics.
expressing the equation of state as a total differential of V̄(T, P):
In its simplest form, first stated by Rudolf Clausius in the
dV̄ = (∂V̄/∂T)P dT + (∂V̄/∂P)T dP = (R/P)dT − (RT/P2)dP, middle of the nineteenth century, the Second Law says
and substitute the result into w1 and w2 above. The result, after that heat cannot spontaneously pass from a cool body to
integration, is that the work performed on path 1 is: a hotter one. Another way of stating this, which may also
be useful, is that any natural process involving a transfer
w1 = R[(T2 − T1) + T1 ln(P1 /P2)]
of energy is inefficient, with the result that a certain
but the work performed on path 2 is: amount is irreversibly converted into heat that cannot be
w2 = R[(T2 − T1) + T2 ln(P1 /P2)], involved in further exchanges. The Second Law, there-
fore, is a recognition that natural processes have a sense
which is clearly different. Similarly, if we had a function for dq,
the heat gained, we could integrate it to show that q1 and q2, the
of direction.
amounts of heat expended along these paths, must also be dif- Two difficult concepts are embedded in what we have
ferent. We are about to do just that. just said. One is spontaneity and the other is reversi-
bility. A spontaneous change is one that, under the right
40 G E O C H E M I S T R Y : PAT H WAY S A N D P R O C E S S E S

if they were reversible. The conceptual challenge arises

because true reversibility is beyond our experience with
the natural world.
Another perspective on the Second Law, then, is that
it tells us that nature favors spontaneous change, and
that maximum work is never performed in real-world
processes. Just as the internal energy function was intro-
duced to present the First Law in a quantitative fashion,
we need to define a thermodynamic function that quan-
tifies the sense of direction and systematic inefficiency that
the Second Law identifies in natural processes. This new
function, known as entropy and given the symbol S, is
defined in differential form by:

dS = (dq/T)rev, (3.3)

in which dq is an infinitesimal amount of heat gained by

a body at temperature T in a reversible process. It is a
FIG. 3.4. Two weights are connected by a rope that passes over
a frictionless pulley. If one weight is infinitesimally heavier than measure of the degree to which a system has lost heat
the other, then it will perform the maximum possible amount of and therefore some of its capacity to do work. Many
work (the work of lifting the other weight) as it falls. The smaller people find entropy an elusive concept to master, so it is
the weight difference between them is, the greater the amount of best to become familiar with the idea by discussing ex-
work and the more nearly reversible it is. amples of its use and properties.
Figure 3.5 illustrates a potentially reversible path for
a system enclosed in a nonadiabatic wall. Suppose that
conditions, can be made to perform work. The heavy the system expands isothermally from state A to state
weight in figure 3.4 can perform the work of raising the B. In doing so, it performs an amount of work on its
lighter weight as it falls, so that this change is sponta- surroundings that can be calculated by the integral of
neous. If the lighter weight raised the heavier one, we PdV. Graphically, this integral is represented by the area
would recognize the change as nonspontaneous and we AA′B′B. Because this is an isothermal process, state A
would assume that some change outside the system had and state B must be in equilibrium with each other,
probably caused it. When we face a conceptual challenge
in telling whether a change is spontaneous, it is usually
because we haven’t defined the bounds of the system
clearly. Notice also that a spontaneous change doesn’t
have to perform work; it just has to be capable of it. The
heavy weight will fall spontaneously whether it lifts the
lighter weight or not.
This leads us to consider reversibility. A change is
reversible if it does the maximum amount of possible
work. The falling weight in figure 3.4, for example, will
be undergoing a reversible change if it lifts an identical
weight and loses no energy to friction in the pulley and
rope that connects it to the other weight. Experience tells
us that reversibility is an unattainable ideal, of course,
FIG. 3.5. The work performed by isothermal compression from
but that doesn’t stop engineers from designing counter- state A to state B is equal to the area AA′B′B. Because dE = 0, an
weighted elevators to use as little extra energy as possible. equivalent negative amount of heat is gained by the system. If
It also doesn’t stop geochemists from considering gradual this change could be reversed precisely, the net change in entropy
changes in systems that are never far from equilibrium as dS would be zero.
 A First Look at Thermodynamic Equilibrium 41

Remember, though, that the Second Law tells us that heat

can only pass spontaneously from a hot body to a cooler
one, so this result is only valid if T2 > T1. Therefore, the
entropy of an isolated system can only increase as it
approaches internal equilibrium:

dSsys > 0.

Finally, look at the closed system illustrated in fig-

ure 3.7. It is similar to the previous system in all respects,
except that is bounded by a nonadiabatic wall, so that
the two bodies can exchange amounts of heat dq1′ and
dq2′ with the world outside. As they approach thermal
equilibrium, the heat exchanged by each is given by:
FIG. 3.6. Two bodies in an isolated system are separated by a
nonadiabatic wall and may, therefore, exchange quantities of heat (dq1)total = dq1 + dq1′,
dq1 and dq2 as they approach thermal equilibrium.
and

(dq2)total = dq2 + dq2′.

which is another way of saying that the internal energy
Any heat exchanged internally must still show up either
of the system must remain constant along the path be-
in one body or the other, so as before:
tween them. Therefore, the work performed by the sys-
tem must be balanced by an equivalent amount of heat dq1 = −dq2.
gained, according to equation 3.2. Suppose, now, that it
The net change in system entropy is, therefore,
were possible to compress the system isothermally, thus
reversing along the path B → A without any interference dSsys = (dq1)total + (dq2)total
due to friction or other real-world forces. We would find
= (dq1′/T1) + (dq2′/T2) + dq1([1/T1] − [1/T2]).
that the work performed on the system (the heat lost by
the system) would be numerically equal to the work on It has already been shown that the Second Law requires
the forward path, but with a negative sign. That is, the the last term of this expression to be positive. Therefore,
net change in entropy around the closed path is:
dSsys > (dq1′/T1) + (dq2′/T2).
dS = dq/T − dq/T = 0.

The entropy function for a system following a reversible

pathway, therefore, is a function of state, just like the
internal energy function.
Next, consider an isolated system in which there are
two bodies separated by a nonadiabatic wall (fig. 3.6).
Let the two bodies initially be at different temperatures
T1 and T2. If we allow them to approach thermal equi-
librium, there must be a transfer of heat dq2 from the
body at temperature T2. Because the total system is iso-
lated, an equal amount of heat dq1 must be gained by the
body at temperature T1. The net change in entropy for
the system is:

dSsys = dS1 + dS2,

or FIG. 3.7. The system in figure 3.6 is allowed to exchange heat

with its surroundings. The net change in internal entropy depends
dSsys = dq1 ([1/T1] − [1/T2]). on the values of dq1′ and dq2′.
42 G E O C H E M I S T R Y : PAT H WAY S A N D P R O C E S S E S

This result is more open ended than the previous one. the other with 1 mole of argon. If the wall is removed, we ex-
Although the entropy change due to internal heat ex- pect that the two gases will mix spontaneously rather than re-
changes in any system will always be positive, it says maining in their separate ends of the system. It would require
considerable work to separate the nitrogen and argon again.
that we have no way of predicting whether the overall
(See problem 3.5 at the end of this chapter.) By randomizing the
dSsys of a nonisolated system will be positive unless we
positions of gas molecules and contributing to a loss of order
also have information about dq1′, dq2′, and the tempera- in the system, we have caused an increase in entropy.
ture in the world outside the system. Suppose that the mixing takes place isothermally, and that
To appreciate this ambiguity, think back to the differ- the only work involved is mechanical. Assume also, for the sake
ence between placing a glass of ice water on the kitchen of simplicity, that both argon and nitrogen are ideal gases. How
counter and placing it in the freezer. In both cases, inter- much does the entropy increase in the mixing process? First,
write equation 3.2 as:
nal heat exchange results in an increase in system en-
tropy. When the glass is refrigerated, however, (dq1′/T1) dq = dE + PdV.
and (dq2′/T2) become potentially large negative quantities, Because we agreed to carry out the experiment at constant
with the result that the change due to internal processes temperature, dE must be equal to zero. Using the equation of
is overwhelmed and the entropy within the glass de- state for an ideal gas, we can see that dq now becomes:
creases. The Second Law, therefore, does not rule out
dq = (nRT/V)dV,
the possibility that thermodynamic processes can reverse
direction. It does tell us, though, that a spontaneous from which:

change can only reverse direction if heat is lost to some dS = dq/T = (nR/V)dV.
external system. A local decrease in entropy must be ac- (Notice that this problem is cast in terms of volume, rather than
companied by an even larger increase in entropy in the molar volume, because we need to keep track of the number of
world at large. When we have difficulty with this notion, moles, n.) We can think of the current problem as one in which
as we commented earlier, it is usually because we have each of the two gases has been allowed to expand from its ini-
not been clear about defining the size and boundaries of tial volume (V1 or V2 ) into the larger volume V1 + V2 . For each,
the system. Therefore, except for those cases in which a then, the entropy change due to isothermal expansion is:
system evolves sufficiently slowly that we are justified in ∆Si = ∆ni R ln([V1 + V2 ]/Vi ),
approximating its path as a reversible one, its change in
and the combined entropy change is:
entropy must be defined not by equation 3.3, but by:
∆S1 + ∆S2 = n1R ln([V1 + V2 ]/V1) + n2 R ln([V1 + V2 ]/V2 ).
dS > (dq/T )irrev. (3.4) This can be simplified by defining the mole fraction, Xi, of either
gas as Xi = ni /(n1 + n2) = Vi /(V1 + V2 ); thus,

Entropy and Disorder ∆S̄ = −R(X1 ln X1 + X2 ln X2),

where the bar over ∆S̄ is our standard indication that it has
Entropy is commonly described in elementary texts as been normalized by n1 + n2 and is now a molar quantity. The
a measure of the “disorder” in a system. For those who entropy of mixing will always be a positive quantity, because X1
learned about entropy through statistical mechanics, fol- and X2 are always <1.0. The answer to our question, therefore,
lowing the approach pioneered by Nobel physicist Max is that entropy increases by:
Planck in the early 1900s, this may make sense. This is 2 × ∆S̄ = 2 × −(1.987)(0.5 ln 0.5 + 0.5 ln 0.5)
sometimes less than satisfying for people who address
= 2.76 cal K−1 or 11.55 J K−1.
thermodynamics as we have, because the mathematical
definition in equations 3.3 and 3.4 doesn’t seem to address
entropy in those terms. To see why it is valid to think of en-
tropy as an expression of disorder, consider this problem. REPRISE: THE INTERNAL ENERGY
FUNCTION MADE USEFUL
Worked Problem 3.2
With the First and Second laws in hand, we can now
Two adjacent containers are separated by a removable wall. return to the subject of equilibrium, which was defined
Into one of them, we introduce 1 mole of nitrogen gas. We fill informally at the beginning of this chapter as a state in
 A First Look at Thermodynamic Equilibrium 43

ENTROPY AND DISORDER: WORDS OF CAUTION

The description of entropy as a measure of disorder dynamics. The biologic concept of order, although
is probably the most often used and abused concept potentially connected to thermodynamics in the realm
in thermodynamics. In many applications, the idea of biochemistry, is based largely on functional com-
of “disorder” is coupled with the inference from the plexity, not equations for energy utilization. The sit-
Second Law that entropy should increase through uation becomes even murkier when applied to groups
time. Like all brief descriptions of complex ideas, of organisms, rather than to individuals.
though, the statement that “entropy measures system Furthermore, from a thermodynamic perspective,
disorder” must be treated with some care. It is com- living organisms cannot be viewed as isolated systems,
mon practice among chemical physicists to estimate separated from their inorganic surroundings. As we
the entropy of a system by calculating its statistical have tried to emphasize, the definition of system
degrees of freedom—computing, in effect, the num- boundaries is crucial if we are to interpret system
ber of different ways that atoms can be configured in changes by the Second Law. A refrigerator, for ex-
the system, given its bulk conditions of temperature, ample, might be mistakenly seen to violate the Sec-
pressure, and other intensive parameters. A less rig- ond Law if we failed to recognize that entropy in the
orous description of disorder, however, can lead to kitchen around it increases even as the contents of
very misleading conclusions about the entropy of a the refrigerator become more ordered.
system and thus the course of its evolution. Finally, we note that there may be instances in
Creationists, for example, have grasped at the idea which entropy and disorder in the macroscopic sense
of entropy as disorder as a vindication of their belief can be decoupled, even in a properly defined isolated
that biologic evolution is impossible. In their view, system. For example, if two liquids are mixed in an
“higher” organisms are more ordered than the more adiabatic container, we might expect that they will
primitive organisms from which biologists presume mix to a random state at equilibrium. We know,
that they evolved. Because this interpretation presents however, that other system properties may commonly
evolution as a historical progression from a less or- preclude a random mixture. Oil and vinegar in a salad
dered to a more ordered state, it describes an apparent dressing unmix readily, as a manifestation of differ-
violation of the Second Law. There are so many flaws ences in their bonding properties. If the attractive force
in this argument that it is difficult to know where to between similar ions or molecules is greater than that
begin to discuss it, and we will mention only a few. between dissimilar ones, then complete random mix-
The first is the question of the meaning of “order” ing would imply an increase, rather than a decrease, in
as applied to organisms or families of organisms. This both internal energy and entropy. A macroscopically
is far from a semantic problem, because it is by no ordered system, in other words, can easily be more
means clear that higher organisms are more ordered stable than a disordered one without shaking our
than primitive ones, from the point of view of thermo- faith in the Second Law of Thermodynamics.

which no change is taking place. The First Law restates Under most geologic conditions, mechanical (pressure-
this condition as one in which dE is equal to zero. From volume) work is the only significant contribution to dw,
the Second Law, we find that entropy is always maxi- and it is reasonable to substitute PdV for dw:
mized during the approach to equilibrium, and that dS
becomes equal to zero once equilibrium is reached. dE ≤ TdS − PdV. (3.5)
The results expressed in equations 3.3 and 3.4 can be
used to recast equation 3.2 as: This is a practical form of equation 3.2, although only
correct under the conditions just discussed. In most geo-
dE ≤ TdS − dw. logic environments, we assume that change takes place
44 G E O C H E M I S T R Y : PAT H WAY S A N D P R O C E S S E S

very slowly, so that systems can be regarded as following Enthalpy

nearly reversible paths. Generally, therefore, the inequal-
ity in equation 3.5 can be disregarded, even though it is The first of these equations can be derived by writing
strictly necessary. equation 3.5 as:
Because much of the remaining discussion of ther-
dE + PdV ≤ TdS.
modynamics in this book derives from equation 3.5, we
should recognize three other relationships that follow If we restrict ourselves by looking only at processes
directly from it and from the fact that the internal energy that take place at constant pressure, this is the same as
function is a function of state. First, equation 3.5 is a dif- writing:
ferential form of E = E(S, V). Assuming that changes take d(E + PV) ≤ TdS,
place reversibly, it can be written as a total differential:
because d(PV) = PdV + VdP, which equals PdV if dP = 0.
dE(S, V) = (∂E/∂S)V dS + (∂E/∂V)S dV. The quantity E + PV is a new function, called en-
Comparison with equation 3.5 yields the two statements: thalpy and commonly given the symbol H. It is useful as
a measure of heat exchanged under isobaric conditions,
T = (∂E/∂S)V , where dH = dq. Reactions that evolve heat, and therefore
and have a negative change in enthalpy, are exothermic. Those
that result in an increase in enthalpy are endothermic.
P = −(∂E/∂V)S . The left side of this last equation can be expanded to
In other words, the familiar variables temperature reveal a differential form for dH:
and pressure can be seen as expressions of the manner dH = d(E + PV) = dE + PdV + VdP,
in which the internal energy of a system responds to
changes in entropy (under constant volume conditions) or
or volume (under adiabatic conditions). Second, because
E(S, V) is a function of state, dE(S, V) is a perfect differ- dH ≤ TdS + VdP. (3.7)
ential (perfect differentials are reviewed in appendix A).
As we did with equation 3.5, we can compare the total
That is,
differential dH(S, P) and equation 3.7 under reversible
(∂2E/∂S∂V) = (∂2E/∂V∂S), conditions to recognize that:
or T = (∂H/∂S)P,

and
(∂T/∂V)S = −(∂P/∂S)V. (3.6)
V = (∂H/∂P)S.
This last equation is known as one of the Maxwell rela-
Enthalpy, like internal energy and entropy, can be
tionships. These and other similar expressions to be
shown to be a function of state, because it experiences
developed shortly can be used to investigate the inter-
no net change during a reversible cycle of reaction paths.
actions of S, P, V, and T. Some of these will be discussed
This makes it possible, among other things, to determine
more fully in chapter 9, when we look in more detail at
the amount of heat that would be exchanged in geolog-
the effects of changing temperature or pressure in the
ically important reactions, even when those reactions may
geologic environment.
be too sluggish to be studied directly at the low tempera-
tures at which they take place in nature. The careful de-
AUXILIARY FUNCTIONS OF STATE termination of such values is the business of calorimetry.
Because H is a function of state, it is also possible to
Equation 3.5 is adequate for solving most thermody-
extract one more Maxwell relation like equation 3.6 from
namic problems. In many situations, however, it is pos-
its cross-partial derivatives (also reviewed in appendix A):
sible to use equations of greater practical interest, which
can be derived by imposing environmental constraints
(∂T/∂P)S = (∂V/∂S)P. (3.8)
on the problem.
 A First Look at Thermodynamic Equilibrium 45

The Helmholtz Function The total differential dF(T, V), when compared with
equation 3.9, yields the useful expressions:
By analogy with the way we introduced enthalpy,
S = −(∂F/∂T)V,
we can discover another useful function by writing equa-
tion 3.5 as: and

dE − TdS ≤ −PdV. P = −(∂F/∂V)T .

Under isothermal conditions, this expression is equiva- Because F is a function of state and dF(T, V ) is there-
lent to: fore a perfect differential, we also gain another Maxwell
relation:
d(E − TS) ≤ −PdV.
(∂S/∂V)T = (∂P/∂T)V . (3.10)
The function F = E − TS is best referred to as the Helm-
holtz function, although you may see it referred to else-
Although the Helmholtz function is rarely used in
where as Helmholtz free energy or the work function.
geochemistry, the Maxwell relationship equation 3.10 is
Unfortunately, a variety of symbols have been used
quite widely used. When we return for a second look at
for the internal energy, enthalpy, and Helmholtz func-
thermodynamics in chapter 9, equation 3.10 will be
tions, as well as Gibbs free energy, which is discussed
discussed as the basis of the Clapeyron equation, a means
shortly. This has led to some confusion in the literature.
for describing pressure-temperature relationships in
In the United States, F is commonly used to designate
geochemistry.
Gibbs free energy (for example, in publications of the
National Bureau of Standards). The International Union
Gibbs Free Energy
of Pure and Applied Chemistry, however, has recom-
mended that G be the standard symbol for Gibbs free
The most frequently used thermodynamic quantity
energy. This usage, if not yet standard, is at least wide-
in geochemistry can be derived by writing equation 3.5 in
spread. In the United States, those who use F for Gibbs
the form:
free energy generally use the symbol A for the Helmholtz
function. To complete the confusion, internal energy, dE − TdS + PdV < 0.
which we have identified with the symbol E, is fre- Under conditions in which both temperature and pressure
quently referred to as U, to avoid confusing it with total are held constant, this expression becomes:
(internal plus potential plus kinetic) energy. Always be
sure you know which symbols you are using. Nicolas d(E − TS + PV) < 0.
Vanserg has written an excellent article on the subject, In the now familiar fashion in which we have already
which is listed among the references at the end of this defined H and F, we designate the quantity E − TS + PV
chapter (Vanserg 1958). with the symbol G and call it the Gibbs free energy, in
Because it might be mistaken for Gibbs free energy, honor of the Josiah Willard Gibbs, a chemistry professor
which is ultimately a more useful function in geochem- at Yale University, who wrote a classic series of papers in
istry, it is best to avoid the term Helmholtz free energy. the 1870s in which virtually all of the fundamental equa-
The name work function, however, is fairly informative. tions of modern thermodynamics appeared for the first
The integral of dF at constant temperature is equal to time. Colloquially, among geochemists, it is common to
the work performed on a system. This expression has its speak simply of “free energy.”
greatest application in mechanical engineering. The differential dG can be written:
It is easy to derive the differential form dF:
dG = d(E − TS + PV) = dE − TdS − SdT
dF = d(E − TS) = dE − TdS −SdT, + PdV + VdP,

or or

dF ≤ −SdT − PdV. (3.9) dG ≤ −SdT + VdP. (3.11)

46 G E O C H E M I S T R Y : PAT H WAY S A N D P R O C E S S E S

As with the previous fundamental equations, we can where ∆G = G2 − G1, ∆H = H2 − H1, and ∆S = S2 − S1.
apply our knowledge of the total differential dG(T, P) of Energy is available to produce a spontaneous change
this state function to find that: in a system as long as ∆G is negative. According to equa-
tion 3.13, this can be accomplished under any circum-
−S = (∂G/∂T )P , V = (∂G/∂P)T ,
stances in which ∆H − T∆S is negative. Most exothermic
and changes, therefore, are spontaneous. Endothermic pro-
cesses (those in which ∆H is positive) can also be spon-
−(∂S/∂P)T = (∂V/∂T)P . (3.12) taneous, but only if they are associated with a large
positive change in entropy. This possibility was not ap-
It can be seen from equation 3.11 that the Gibbs free preciated at first by chemists. In fact, in 1879 the French
energy is the first of our fundamental equations written thermodynamicist Marcelin Berthelot used the term
solely in terms of the differentials of intensive parame- affinity, defined by A = −∆H, as a measure of the direc-
ters. This and the ease with which both temperature and tion of positive change. According to his reasoning, a
pressure are usually measured contributes to the great spontaneous chemical reaction could only occur if A > 0.
practical utility of this function. If an endothermic reaction turned out to be spontaneous,
However, what is “free” about the Gibbs free en- he assumed that some unobserved mechanical work
ergy? Consider the intermediate step in equation 3.11. must have been done on the system. As chemists be-
At constant temperature and pressure, this reduces to: came familiar with Gibbs’s papers on thermodynamics,
dG = dE − TdS + PdV. however, it became clear that Berthelot and others had
misunderstood the concept of entropy. Affinity was a
Substituting dE = dq − dw, we have:
theoretical blind alley.
dG = dq − dw − TdS + PdV. In summary, a thermodynamic change proceeds as
long as there can be a further decrease in free energy.
If the quantity of heat dq is transferred to the system
Once free energy has been minimized (that is, when dG
isothermally and any changes are reversible, then dq = TdS
= 0), the system has attained equilibrium.
and dw = dwrev , so we can rewrite this last expression as:
−dG = dwrev − PdV.
Worked Problem 3.3
The decrease in free energy of a system undergoing a
reversible change at constant temperature and pressure, To illustrate how enthalpy, entropy, and temperature are re-
therefore, is equal to the nonmechanical (i.e., not pressure- lated to G, consider what happens when ice or snow sublimates
volume) work that can be done by the system. If the at constant temperature. Those who live in northern climates
condition of reversibility is relaxed, then: will recognize this as a common midwinter phenomenon. Fig-
ure 3.8 is a schematic representation of the way in which vari-
−dG > dwirrev − PdV. ous energy functions change as functions of the proportions of
In either case, the change in G during a process is a meas- water vapor and ice in a closed system. (Strictly speaking, snow
does not sublimate into a closed system, but into the open at-
ure of the portion of the system’s internal energy that is
mosphere. On a still night, close to the ground, however, this
“free” to perform nonmechanical work.
is not a bad approximation.) Notice that the enthalpy of the
The free energy function provides a valuable practical system increases linearly as vaporization takes place. This can
criterion for equilibrium. At constant temperature, the be rationalized by noting we have to add heat to the system to
net change in free energy associated with a change from break molecular bonds in the ice. If enthalpy were the only fac-
state 1 to state 2 of a system can be calculated by inte- tor involved in the process, therefore, the system would be most
grating dG: stable when it is 100% solid, because that is where H is mini-
mized. Entropy, however, is maximized if the system is 100%

∫ dG = ∫ d(E − TS + PV ) = ∫ d(H − TS),

2 2 2
water vapor, because the degree of system randomness is great-
1 1 1
est there. It can be shown from statistical arguments, however,
from which: that entropy is a logarithmic function of the proportion of vapor,
rising most rapidly when the vapor fraction in the system is low.
∆G = ∆H − T∆S, (3.13) Therefore the free energy of the system, G = H − TS, is less than
the free energy of the pure solid until a substantial amount of
 A First Look at Thermodynamic Equilibrium 47

In our discussions so far, we have perhaps given the

impression that whatever thermodynamic data we might
need to solve a particular problem will be readily avail-
able. The references in appendix B do, in fact, include
data for a large number of geologic materials. We have
largely ignored the thorny problem of where they come
from, however, and where we turn for data on materials
that have not yet appeared in the tables. As an example
of how thermodynamic data are acquired, we now show
how solution calorimetry is used to measure enthalpy of
formation. In chapter 10, we show how to extract ther-
modynamic data from phase diagrams.
FIG. 3.8. The quantity G = H − TS varies with the amount of It is not necessary (or even possible) to make a direct
vapor as sublimation takes place in an enclosed container. The determination of ∆Hf0 for most individual phases, be-
free energy when vapor and solid are in equilibrium is Gmin. cause we can rarely generate the phases from their con-
(Modified from Denbigh 1968.)
stituent elements. Instead, we measure the heat evolved
or absorbed as the substance we are interested in is pro-
duced by a specific reaction between other phases for
vapor has been produced. Sublimation occurs spontaneously.
At some intermediate vapor fraction, H − TS reaches a mini- which we already have enthalpy data. Because most re-
mum, and solid and vapor are in equilibrium at Gmin. If the actions of geologic interest are abysmally slow at low
vapor fraction is increased further, the difference H − TS ex- temperatures, even this enthalpy change may be almost
ceeds Gmin again, and condensation occurs spontaneously. This impossible to measure directly. However, because en-
is the source of the beautiful hoar frost that develops on tree thalpy is an extensive variable, we can employ some
branches and other exposed surfaces on still winter mornings. sleight-of-hand: we can measure the heat lost or gained
when the products and reactants are each dissolved in sep-
arate experiments in some solvent at low temperature,
CLEANING UP THE ACT: CONVENTIONS and then add the heats of solution for the various prod-
FOR E, H, F, G, AND S ucts and reactants to obtain an equivalent value for the
heat of reaction.
Except in the abstract sense that we have just used G in
worked problem 3.3, there is no way we can talk about
Worked Problem 3.4
absolute amounts of energy in a system. To say that a
beaker of reagents contains 100 kJ of enthalpy or 55 kcal
Consider the following laboratory exercise. We wish to deter-
of free energy is meaningless. Instead, we compare the
mine the molar enthalpy for the reaction:
value of each of the energy functions to its value in a
2MgO + SiO2 ← → Mg2SiO4,
mixture of pure elements under specified temperature
periclase quartz forsterite
and pressure conditions (a standard state). For example,
we would measure the enthalpy of a quantity of NaCl at to which we assign the value ∆H̄1. This can be determined by
298 K relative to the enthalpies of pure sodium and pure measuring the heats of solution for periclase, quartz, and for-
sterite in HF at some modest temperature:
chlorine at the same temperature and at one atmosphere
of pressure, using the symbol ∆Hf0. The superscript 0 in- → (2MgO, SiO )
2MgO + SiO2 + HF ← 2 solution ,
dicates that this is a standard state value and the sub- which gives ∆H̄2, and:
script f indicates that the reference standard is a mixture
→ (2MgO, SiO )
Mg 2SiO4 + HF ←
of pure elements. By convention, the ∆Hf0 or ∆Gf0 for a 2 solution ,

pure element at any temperature is equal to zero. The which gives ∆H̄3. If the solutions are identical, we can see that
same notation is used for internal energy and the Helm- the first of these equations is mathematically equivalent to the
holtz function as well, although they are less frequently second minus the third, so that:

encountered in geochemistry. ∆H̄1 = ∆H̄2 − ∆H̄3.

48 G E O C H E M I S T R Y : PAT H WAY S A N D P R O C E S S E S

In this way, solution calorimetry can be used to determine the With this new perspective, return for a moment to
enthalpy of formation for forsterite from its constituent oxides equation 3.13. It should now be clear that the symbol ∆
at low temperature. As you might expect, however, solvents has a different meaning in this context. In fact, although
suitable for silicate minerals (such as hydrofluoric acid or molten
it is rarely done, it would be less confusing to write equa-
lead borate) are generally corrosive and hazardous to handle.
tion 3.13 as:
Consequently, these experiments require considerable skill and
specialized equipment. Most geochemists refer to published ∆(∆G) = ∆(∆H) − T∆S,
collections of calorimetric data rather than making the meas-
urements themselves. in which the deltas outside the parentheses and on S refer
What we have just described is not a measurement of ∆H̄f0, to a change in state (that is, a change in these values dur-
because the reference comparison was to constituent oxides, ing some reaction), and the deltas inside the parentheses
not pure elements. If we wanted to determine ∆H̄f0, we would refer to values of G or H relative to some reference state.
search for tabulated data for reactions forming the oxides from We examine this concept more fully in later chapters.
pure elements (typically measured by some method other than
Although we have not felt it necessary to prove, it
solution calorimetry). Again, recalling that enthalpies are addi-
should be apparent that each of the functions E, H, F, G,
tive, we could recognize that ∆H̄f0 for MgO is defined by the
reaction: and S is an extensive property of a system. The amount of
energy or entropy in a system, therefore, depends on the
→ 2MgO,
2Mg + O2 ←
size of the system. In most cases, this is an unfortunate
and ∆H̄f0 for SiO2 is defined by: restriction, because we either don’t know or don’t care
→ SiO .
Si + O2 ← how large a natural system may be. For this reason, it is
2
customary to normalize each of the functions by divid-
Label these two values ∆H̄4 and ∆H̄5. Therefore, for the en-
ing them by the total number of moles of material in the
thalpy ∆H̄6 of the net reaction:
system, thus making each of them an intensive property.
→ Mg SiO ,
2Mg + 2 O2 + Si ← 2 4

we obtain: COMPOSITION AS A VARIABLE

∆H̄6 = ∆H̄2 − ∆H̄3 + ∆H̄4 + ∆H̄5.
Up to now, the functions we have considered all assume
The value ∆H̄6 in this case is the enthalpy of formation for for- that a system is chemically homogeneous. Most systems
sterite derived from those for the elements, ∆H̄f0. of geochemical interest, however, consist of more than
one phase. The problem most commonly faced by geo-
chemists is that the bulk composition of a system can be
Although most thermodynamic functions are best packaged in a very large number of ways, so that it is
defined as relative quantities like ∆Hf0, entropy is a major generally impossible to tell by inspection whether the as-
exception. The convention observed most commonly is semblage of phases actually found in a system is the most
an outgrowth of the Third Law of Thermodynamics, likely one. To answer questions dealing with the stabil-
which can be stated, in paraphrase from Lewis and Ran- ity of multiphase systems, we need to write a separate set
dall (1961): If the entropy of each element in a perfect of equations for E, H, F, and G for each phase and apply
crystalline state is defined as zero at the absolute zero of criteria for solving them simultaneously. We will do this
temperature, then every substance has a nonnegative job in two steps.
entropy; at absolute zero, the entropy of all perfect crys-
talline substances becomes zero. This statement, which
Components
has been tested in a very large number of experiments,
To describe the possible variations in the composi-
provides the rationale for choosing the absolute temper-
tions and proportions of phases, it is necessary to define
ature scale (i.e., temperature in Kelvins) as our standard
a set of thermodynamic components that satisfy the fol-
for scientific use. For now, the significance of the Third
lowing rules:
Law is that it defines a state in which the absolute or
third law entropy is zero. Because this state is the same 1. The set of components must be sufficient to describe
for all materials, it makes sense to speak of S0, rather all of the compositional variations allowable in the
than ∆Sf0. system.
 A First Look at Thermodynamic Equilibrium 49

2. Each of the components must vary independently in

the system.

As long as these criteria are met, the specific set of

components chosen to describe a system is arbitrary,
although there may often be practical reasons for choos-
ing one set rather than another.
These rules set very stringent restrictions on the way FIG. 3.9. Olivine and pyroxene solid solutions can each be repre-
components can be chosen, so it is a good idea to spend sented by two end-member components.
time examining them carefully. First, notice that the solid
phases we encounter most often in geochemical situations olivine’s composition varies only between Mg2SiO4 (Fo) and
have compositions that are either fixed or are variable Fe2SiO4 (Fa), and that orthopyroxene is a solid solution be-
only within bounds allowed by stoichiometry and crystal tween MgSiO3 (En) and FeSiO3 (Fs). What components might
structure. This means that if we are asked to find com- be used to describe olivine and orthopyroxene individually,
ponents for a single mineral, they must be defined in such and how might we select components for an ultramafic rock
a way that they can be added to or subtracted from the consisting of both minerals?
The simplest mineral components are the end-member
mineral without destroying its identity. A system con-
compositions themselves. The end-member compositions are
sisting only of rhombohedral Ca-Mg carbonates, for completely independent of one another in each mineral and, as
example, cannot be described by entities such as Ca2+, can be seen at a glance in figure 3.9, any mineral composition
MgO, or CO2, because none of them can be independ- in either solid solution can be formed by some linear combina-
ently added to or subtracted from the system without vi- tion of the end members. Compositions corresponding to Fo,
olating its crystal chemistry. The most obvious, although Fa, En, and Fs, therefore, satisfy our selection rules.
not unique, choice of components in this case would be A choice of FeO or MgO would not be valid, because
neither one can be added to or subtracted from olivine or or-
CaCO3 and MgCO3.
thopyroxene unless we also add or subtract a stoichiometric
When a system consists of more than one phase, it is
amount of SiO2. Changing FeO or MgO alone would produce
common to find that some components selected for in-
dividual phases are redundant in the system as a whole.
This occurs because it is possible to write stoichiometric
relationships that express a component in one phase as
some combination of those in other phases. For each
stoichiometric equation, therefore, it is possible to re-
move one phase component from the list of system com-
ponents and thus to arrive at the independent variables
required by rule 2. It is also possible, and often desirable,
to choose system components that cannot serve as com-
ponents for any of the individual phases in isolation. Such
a choice is compatible with the selection rules if the
amount of the component in the system as a whole can
be varied by changing proportions of individual phases.
Petrologists usually refer to nonaluminous pyroxenes, for
example, in terms of the components MgSiO3, FeSiO3,
and CaSiO3, even though CaSiO3 cannot serve as a com-
ponent for any pyroxene considered by itself.

Worked Problem 3.5 FIG. 3.10. If we choose to describe a system containing olivines
and pyroxenes, we need three components. Phase compositions
Olivine and orthopyroxene are both common minerals in basic lying outside the triangle of system components require negative
igneous rocks. For the purposes of this problem, assume that amounts of one or more components.
50 G E O C H E M I S T R Y : PAT H WAY S A N D P R O C E S S E S

phase with the composition FeSiO3. Monatomic compo-

nents such as F, S, or O are also legitimate, even though
fluorine, sulfur, and oxygen invariably occur as molecules
containing two or more atoms. For some petrological
applications, it makes sense to use such components as
CaMg−1, which clearly do not exist as real substances. In
fact, the carbonates discussed above can be characterized
quite well by MgCO3 and CaMg −1, as can be seen from
the stoichiometric relationship:

MgCO3 + CaMg −1 = CaCO3.

Components of this type, known as exchange operators,

have been used to great advantage in describing many
metamorphic rocks (Thompson et al. 1982; Ferry 1982).
FIG. 3.11. An alternative selection of components for the system
in figure 3.10. All olivine and pyroxene compositions now lie
Because components are abstract constructions, we are
within the triangle. not required to use them in positive amounts. Fe2O3, for
example, can be described by the components Fe3O4 and
Fe, even though we need to add a negative amount of
system compositions that lie off of the solid solution lines in Fe to Fe3O4 to do the job:
figure 3.9.
If olivine and orthopyroxene are not isolated phases but are 3Fe3O4 − Fe = 4Fe2O3.
constituents of a rock, however, we need to choose a different
set of components. The mineral components are still appropri-
ate, but one of them is now redundant. We can eliminate it by Worked Problem 3.6
writing a stoichiometric relationship involving the other three.
For example: The ratio K/Na in coexisting pairs of alkali feldspars and alkali
MgSiO3 = 0.5Mg2 SiO4 + FeSiO3 − 0.5Fe2 SiO4. dioctahedral micas can be used to infer pressure and tempera-
ture conditions for rocks in which they were formed. We see
Notice that this is only a mathematical relationship among how this is done in chapter 9. For now, let’s ask what selection
abstract quantities, not necessarily a chemical reaction among of components might be most helpful if we were interested in
phases. We have tried to emphasize this by using an equal sign K← → Na exchange reactions between these two minerals.
rather than an arrow. The solid triangle in figure 3.10 illustrates The exchange operator KNa−1 is a good choice for a system
the system as defined by the components on the right side. component in this case. Feldspar compositions can be generated
As required by the negative amount of Fe2SiO4 in the from:
equation above, magnesium-bearing orthopyroxenes have com-
positions outside of the triangle defined by the system compo- NaAlSi3O8 + x KNa−1 = Kx Na1−x AlSi3O8 ,
nents. There is nothing wrong with this representation, but it is for any value of x between 0 and 1. Similarly, mica compositions
awkward for most petrologic applications. A more conven- can be derived from:
tional selection of components is shown in figure 3.11. The
mineral compositions at the ends of the olivine solid solution NaAl3Si3O10(OH)2 + yKNa −1
are still retained as system components, but the third compo- = Ky Na1−yAl3Si3O10 (OH)2.
nent, SiO2, does not correspond to either mineral in the system. If we also select the two sodium end-member compositions as
components, then all possible compositional variations in the
system can be described. Figure 3.12a shows one way of illus-
It is very important to recognize that components are trating this selection. A more subtle diagram, using the same set
of components, is presented in figure 3.12b.
an abstract means of characterizing a system. They do not
This is not a contrived example. We have selected it to em-
need to correspond to substances that can be found in na-
phasize that chemical components are abstract mathematical
ture or manufactured in a laboratory. Orthopyroxenes, entities, but it is also meant to illustrate how a very common
for example, are frequently described by the components class of geochemical problems can be reconceived and simpli-
MgSiO3 and FeSiO3, even though there is no natural fied by choosing components creatively.
 A First Look at Thermodynamic Equilibrium 51

Further infinitesimal changes in the amounts of forsterite

or fayalite in the system would result in a small change
in E:

dE = ĒFo dn Fo + ĒFa dn Fa.

This differential equation can be written in the form of

a total differential, from which it can be seen that:

ĒFo = (∂E/∂n Fo )n Fa,

and

ĒFa = (∂E/∂n Fa )n Fo.

This is an idealized process, of course, chosen to demon-

strate that the internal energy of a phase can be changed—
in addition to the ways we have already discussed—by
varying its composition. It is more realistic to recognize
that the mixing process as described involves increases
in both entropy and volume. Notice also that the total
internal energy of the phase, E, is not a molar quantity,
because it has not been divided by the total number of
moles in the system. The relationships dEolivine = ĒFodn or
dEolivine = Ē Fa dn can only be valid if the olivine is either
pure forsterite or pure fayalite. In between, as we see in
later chapters, dE takes a nonlinear and generally com-
plicated form. The quantities labeled Ē Fo and Ē Fa above
are more useful if written with these restrictions in mind:

Ei = (∂E/∂ni)S,V,n j ≠ i. (3.14)
FIG. 3.12. (a) Alkali micas and feldspars can be described by a
component set that includes KNa−1. (b) An orthogonal version of We have now identified a new thermodynamic quan-
the diagram in (a). Its vertical edges both point to KNa−1. tity, the partial molar internal energy, which describes
the way in which total internal energy for a phase re-
sponds to a change in the amount of component i in the
phase, all other quantities being equal. You may think of
CHANGES IN E, H, F, AND G it, if you like, as a chemical “pressure” or a force for en-
DUE TO COMPOSITION ergy change in response to composition, in the same way
that pressure is a force for energy change in response to
Consider an open system containing only magnesian volume. To emphasize the importance of this new func-
olivine, a pure phase. The internal energy of the system tion, it has been given the symbol µ i and is called the
is equal to E = ĒFonFo, where ĒFo is the molar internal en- chemical potential of component i in the phase.
ergy for pure forsterite and n Fo is the number of moles of The internal energy of a phase, then, is:
the forsterite component in the system. Suppose that it
E = E(S, V, n1, n2, . . , nj).
were possible to add a certain number of moles of the
fayalite component, nFa, to the one-phase system without We can rewrite equation 3.5 to include the newfound
causing any increase in energy as a result of the mixing chemical potential terms:
itself. Because E is an extensive property, the internal en-
ergy of the phase would then be equal to: dE ≤ TdS − PdV + µ1dn1 + µ2dn2 + . . .

E = ĒFo n Fo + ĒFa nFa.

+ µjdnj ≤ TdS − PdV + µidni.Σ (3.15a)
52 G E O C H E M I S T R Y : PAT H WAY S A N D P R O C E S S E S

In the same way, it can be shown that the auxiliary func- It is easiest to examine the equilibrium condition
tions H, F, and G are also functions of composition: among phases if we consider only the simple (and geo-
chemically unlikely) situation in which system compo-
dH ≤ TdS + VdP + µ1dn1 + µ2dn2 + . . . nents correspond one-for-one with components of the
+ µj dnj ≤ TdS + VdP + µi dni ,Σ (3.15b) individual phases. We also consider only the equilibrium
conditions for an isolated system. We are doing this only
dF ≤ −SdT − PdV + µ1dn1 + µ2dn2 + . . .
for the sake of simplicity, however. We would arrive at
+ µj dnj ≤ −SdT − PdV + µi dni , Σ (3.15c)
the same conclusions if we took on the more difficult
dG ≤ −SdT + VdP + µ1dn1 + µ2dn2 + . . . challenge of a closed or open system, or if we considered
+ µj dnj ≤ −SdT + VdP + µidni . Σ(3.15d) a set of system components that differ from the set of all
phase components.
Chemical potential, therefore, can be defined in several Because the system is isolated, we know that its ex-
equivalent ways: tensive parameters must be fixed. That is,

µi = (∂E/∂ni)S,V,n j ≠ i (3.16a)
dEsys = ΣdE Φ =0

dSsys = ΣdS
Φ
=0
= (∂H/∂ni)S,P,n j ≠ i (3.16b)
dVsys = ΣdV
Φ
=0
= (∂F/∂ni)T,V,n j ≠ i (3.16c)
dn1,sys = Σdn =0
= (∂G/∂ni)T,P,n j ≠ i (3.16d) 1,Φ

dn2,sys = Σdn 2,Φ

=0
and can be correctly described as the partial molar en- .
.
thalpy, the partial molar Helmholtz function, or the par- .
tial molar free energy, provided that the proper variables dni,sys = Σdn i,Φ = 0.
are held constant, as indicated in equations 3.16a–d. There is one dni,sys equation for each component in the
system. Despite these equations, there is no constraint
CONDITIONS FOR that prevents the relative values of E, S, V, or the various
HETEROGENEOUS EQUILIBRIUM n’s from readjusting themselves, as long as their totals
remain zero. That is, the various individual values of EΦ,
We are now within reach of a fundamental goal for this VΦ, and the ni,Φ are not independent. Whatever leaves
chapter. Having examined the various ways in which in- one phase in the system must show up in at least one of
ternal energy is affected by changes in temperature, pres- the others. On the other hand, there are constraints on
sure, or composition, we may now ask: what conditions the intensive parameters in the system. To see what they
must be met if a system containing several phases is in are, let’s write an equation for the total internal energy
internal equilibrium? This is a circumstance usually re- change for the system, dEsys:
ferred to as heterogeneous equilibrium.
A system consisting of several phases can be char-
dEsys = Σ dE = 0 = Σ(T dS
Φ Φ Φ) − Σ(P dV )
Φ Φ
+ Σ(Σ (µ dn )).i,Φ iΦ
acterized by writing an equation in the form of equa-
tion 3.15a for each individual phase: The only way to guarantee that Σ dEsys = 0 is for each
of the terms on the right side of this equation to equal
dEΦ ≤ TΦdSΦ − PΦdVΦ + Σ (µ iΦ
dniΦ ),
zero. We have already agreed that dS, dV, and each of
where we are using the subscript Φ to identify properties the dni might differ from phase to phase. It would be a re-
with the individual phase. For example, if the system con- markable coincidence if TΦ, PΦ, and each of the µiΦ could
tained phases A, B, and C, we would write an equation also vary among phases in such a way that Σ(TΦdSΦ),
for dEA, another for dE B, and a third for dEC . The final Σ(PΦdVΦ), and Σ(Σ(µiΦ dniΦ)) were always equal to
term of each equation is the sum of the products µi dni zero. Fortunately, we do not need to rely on coincidence.
for each system component (1 through i) for the individ- Unlike extensive properties, intensive properties are not
ual phase. free to vary among phases in equilibrium. We showed
 A First Look at Thermodynamic Equilibrium 53

earlier in this chapter that this is true for temperature respect to the mole fraction of FeSiO3 is identical in pyroxene
and pressure; at equilibrium and in quartz. If it were possible to add the same infinitesimal
amount of FeSiO3 to each, the free energies of the two phases
TA = TB = . . . = TΦ would each change by the same amount:

PA = PB = . . . = PΦ. dG = µFeSiO dnFeSiO .

3 3

It should be evident now that the same is true for chem-

ical potentials; at equilibrium
The Gibbs-Duhem Equation
µ1A = µ1B = . . . = µ1Φ

µ2A = µ2B = . . . = µ2Φ One final, very useful relationship can be derived from
. this discussion of chemical potentials. Consider equa-
.
. tion 3.15a for a phase that is in equilibrium with other
µiA = µiB = . . . = µiΦ. phases around it. It is possible to write an integrated
form of equation 3.15a:
Because this conclusion is so crucial in geochemistry, we
emphasize it again: At equilibrium, the chemical poten- E = TS − PV + Σµ n . i i
tial of any component must be the same in all phases in a
Therefore, the differential energy change, dE, that takes
system. To be sure that these general results are clear, let’s
place if the system is allowed to leave its equilibrium
examine heterogeneous equilibrium in a specific system.
state by making small changes in any intensive or exten-
sive properties is:
Worked Problem 3.7
dE = TdS + SdT − PdV − VdP + Σ µ dn
i i

An experimental igneous petrologist, working in the laboratory, Σ

+ ni dµi .
has produced a run product which consists of quartz (qz) and To see how the intensive properties alone are interrelated,
a pyroxene (cpx) intermediate in composition between FeSiO3
subtract equation 3.15a, which was written in terms of
and MgSiO3. Assuming that the two minerals were formed in
equilibrium, what conditions must have been satisfied?
variations in extensive parameters alone, from this equa-
To answer this question, first choose a set of components tion to get:
for the system. Several selections are possible, some of which
we considered in an earlier problem. This time, let’s choose the
end-member mineral compositions FeSiO3, MgSiO3, and SiO2.
0 = SdT − VdP + Σ n dµ .
i i
(3.17)

Heterogeneous equilibrium then requires that the intensive pa- This expression is known as the Gibbs-Duhem
rameters be:
equation. It tells us that there can be no net gradients in
Tqz = Tcpx = T intensive parameters at equilibrium. Of more specific in-
Pqz = Pcpx = P terest in geochemical problems, at constant temperature
and pressure, there can be no net gradient in internal
µ SiO ,qz = µ SiO ,cpx = µ SiO
2 2 2 energy as a result of composition in a phase in internal
µ FeSiO ,qz = µ FeSiO ,cpx = µ FeSiO equilibrium. This can be seen, in a way, as another way
3 3 3

µMgSiO ,qz = µMgSiO ,cpx = µMgSiO . of stating the conclusions from our discussion of hetero-
3 3 3
geneous equilibrium.
Notice that the chemical potentials of FeSiO3 and MgSiO3 are
defined in quartz, despite the fact that they are not components
of the mineral quartz itself, and µ SiO2 is defined in pyroxene, SUMMARY
although SiO2 is not a component of pyroxene. All three are
system components. The chemical potential of any component
What is equilibrium? Our discussion of the three laws
is a measure of the way in which the energy of a phase changes
if we change the amount of that component in the phase. For
of thermodynamics has led us to discover several char-
example, if chemical potentials are defined by equation 3.16d, acteristics that answer this question. First, systems in
the constraint on FeSiO3 should be read to mean that at con- equilibrium must be at the same temperature: this is the
stant temperature and pressure, the derivative of free energy with condition of thermal equilibrium. Second, provided that
54 G E O C H E M I S T R Y : PAT H WAY S A N D P R O C E S S E S

work is exclusively defined as the integral of PdV, there enlightening selection of articles designed for classroom use.
can be no pressure gradient between systems in equilib- Chapters 1–4 are particularly relevant to our first discussion
rium: this is the condition of mechanical equilibrium. of thermodynamics. Be prepared to stretch.)
Fraser, D. G., ed. 1977 Thermodynamics in Geology. Dordrecht,
Finally, the chemical potential of each component must
Holland: D. Reidel. (Topical chapters showing both funda-
be the same in all phases at equilibrium. This is the least
mentals and applications of thermodynamics.)
obvious of the conditions we have discussed, and is the Goldstein, M., and I. F. Goldstein. 1978. How We Know: An
focus of many discussions in subsequent chapters. Exploration of the Scientific Process. New York: Plenum.
We have also repackaged the internal energy function (The historical development of our notions of heat and tem-
in several ways and examined the role of entropy, which perature is discussed in chapter 4 of this thought-provoking
will always be maximized during the approach to equi- book.)
Greenwood, H. J., ed. 1977. Short Course in Application of
librium. To demonstrate that each of the equilibrium
Thermodynamics to Petrology and Ore Deposits. Mineral-
criteria is a direct consequence of the three laws, we used
ogical Association of Canada. (A very readable book, with
the internal energy function, E(S, V, n), as the basis of many sections of interest to economic geologists.)
our discussion. As we progress, though, we will quickly Kern, R., and A. Weisbrod. 1967. Thermodynamics for Geolo-
abandon E in favor of the Gibbs free energy. Any of the gists. San Francisco: Freeman, Cooper. (Perhaps the simplest
energy functions, however, can be used to clarify the con- treatment of thermodynamics available for the geologist,
ditions of equilibrium. although now a bit dated.)
With the ideas we have introduced in this chapter, we Lewis, G. N., and M. Randall. 1961. Thermodynamics, revised
by K. S. Pitzer and L. Brewer. New York: McGraw-Hill. (A
are now ready to begin exploring geologic environments.
classic text still in widespread use.)
This has only been a first look at thermodynamics, how- Nordstrom, D. K., and J. Munoz. 1985. Geochemical Thermo-
ever. We return to the topic many times. dynamics. Menlo Park: Benjamin Cummings. (An excellent
text for the serious student.)
suggested readings Smith, E. B. 1982. Basic Chemical Thermodynamics, 3rd ed.
New York: Oxford University Press. (A remarkably clear text
There are many good introductory texts in thermodynamics, for the serious beginner.)
although most are written for chemists rather than for geo- Thompson, J. B., J. Laird, and A. B. Thompson. 1982. Reactions
chemists. Among the best, or at least most widely used, texts for in amphibolite, greenschist, and blueschist. Journal of Petrol-
geochemists are listed here, along with excellent but challeng- ogy 23:1–27. (A good study for the motivated student. Re-
ing articles and historical accounts. actions in amphibole-bearing assemblages are discussed in
terms of exchange operators.)
Denbigh, K. 1968. The Principles of Chemical Equilibrium. Vanserg, N. 1958. Mathmanship. The American Scientist 46:
London: Cambridge University Press. (Heavy going, but al- 94A–98A. (This article does not deal with thermodynamics
most anything you want to know is in here somewhere.) directly, but is a commentary on the use of symbols. Its au-
Ferry, J. M., ed. 1982. Characterization of Metamorphism thor was better known by his real name, Hugh McKinstry,
through Mineral Equilibria. Reviews in Mineralogy 10. professor of geology at Harvard for many years.)
Washington, D.C.: Mineralogical Society of America. (An
 A First Look at Thermodynamic Equilibrium 55

PROBLEMS

(3.1) Refer to figures 3.9 and 3.10. Notice that by adding or subtracting SiO2, we change the proportions
of olivine and pyroxene in a rock, but not their compositions. Changing the amount of any other
system component in figure 3.9 or 3.10, however, alters both the compositions and the proportions
of minerals. Can you find another set of components that includes one which, when varied, changes
the compositions but not the proportions of olivine and pyroxene?

(3.2) Show that the work done by 1 mole of a gas obeying the Van der Waals equation during isothermal
expansion from V1 to V2 is equal to w = RT ln([V̄2 − b]/[V̄1 − b] + a([1/V̄2] − [1/V̄1]).

(3.3) How much work can be obtained from the isothermal, reversible expansion of one mole of chlorine
gas from 1 to 50 liters at 0°C, assuming ideal gas behavior? What if chlorine behaves as a Van der
Waals gas? Use the result of problem 3.2; a = 6.493 l 2 atm mol−2, b = 0.05622 l mol−1.

(3.4) What is the maximum work that can be obtained by expanding 10 grams of helium, an ideal gas,
from 10 to 50 liters at 25°C? Express your answer in (a) calories, (b) joules.

(3.5) Suppose that a mixture of two inert gases is divided into two containers separated by a wall. Imagine
a microscopic valve placed in the wall that allows molecules of one gas to move into container A and
molecules of the other gas to move into container B, but does not allow either gas to move in the
opposite direction. Such an imaginary device is called Maxwell’s demon. How would such a device
violate the Second Law of Thermodynamics?

(3.6) What conditions must be satisfied if hematite, magnetite, and pyrite are in equilibrium in an ore
assemblage?

(3.7) Construct a triangular diagram for the alkali feldspar-mica system similar to figure 3.12a, but in
which you swap the positions of NaAl3Si3O10(OH)2 and KAlSi3O8. What component must now take
the place of KNa−1 at the top corner of the diagram? Where would this component have been plotted
on figure 3.12a? Where does KNa−1 plot on your diagram?

(3.8) Using the Maxwell relations, verify that −(∂V/∂T )P /(∂P/∂T)V = (∂V/∂P)T .