DESY 94-025

February 1994
arXiv:hep-ph/9403219v1 3 Mar 1994

Finite Temperature Effective Potential

to Order g 4, λ2 and
the Electroweak Phase Transition

1
Zoltán Fodor and Arthur Hebecker
Deutsches Elektronen-Synchrotron DESY, Hamburg D-22603, Germany

Abstract

The standard model effective potential is calculated at finite temperature to

order g 4 , λ2 and a complete zero temperature renormalization is performed. In
comparison with lower order calculations the strength of the first order phase
transition has increased dramatically. This effect can be traced back to infrared
contributions from typical non-Abelian diagrams and to the infrared behaviour of
the scalar sector close to the critical temperature. Several quantities, e.g. surface
tension, latent heat and field expectation value are analyzed for an SU(2)-Higgs
model and for the full standard model in detail. An explicit formula enabling
further analytic or numerical study is presented.

1
On leave from Institute for Theoretical Physics, Eötvös University, Budapest, Hungary
1 Introduction
Recently it has been clarified that the electroweak phase transition plays an important role
for the observed baryon asymmetry of the universe [1] (for a recent review see [2]).
Several approaches have been used to determine the details of the electroweak phase
transition. Important results have been obtained by use of 3-dimensional effective theory [3],
ǫ-expansion [4, 5] and average action [6]. There is also a growing interest in lattice simulations
of the phase transition [7].
Perturbative calculations of the finite temperature effective potential of the standard
model have been carried out using the one loop ring summation [8, 9] to order g 3 , λ3/2
(g denotes the gauge-coupling and the top-Yukawa coupling). Two-loop summation has
been done to order g 4 , λ in [10], in which scalar masses have been neglected with respect to
gauge-boson masses, and by use of another approximation in [11].
However, there is a need to extend the work of Arnold and Espinosa [10] to a complete
g , λ2 -calculation. The analysis of the Abelian Higgs model has shown, that higher order λ-
4

corrections can change the g 4 , λ-result for the potential and surface tension significantly [12].
Despite the general scepticism concerning the predictive power of the perturbative results in
the Φ4 -theory, note that calculating far enough in the resummed loop expansion the obtained
effective potential suggests the correct second order phase transition. For the Z2 -symmetric
Φ4 -model one should go to the order λ2 , while for the O(4)-symmetric Φ4 -model even a
λ3/2 -calculation is sufficient.
The calculation of the g 4, λ2 -potential for the standard model and its detailed analysis
are the main goals of the present paper. A systematic expansion in coupling constants is
performed taking into account the effects of infrared divergences [9] and keeping the full
dependence on the Higgs field ϕ, its zero temperature vacuum expectation value v and the
temperature T. The effect of the higher order λ-corrections is found to be important for
realistic Higgs masses.
The principal method of the calculation, based on the Dyson-Schwinger equation for the
derivative ∂V /∂ϕ, is explained in sect. 1 for some general theory containing all the important
features of the standard model. Essentially a summation of tadpole diagrams is performed
[13]. The application to the standard model is carried out and the renormalization at zero
temperature is presented.
In sect. 2 the pure SU(2)-Higgs model is analysed in detail. Using the surface tension
and other physical quantities a comparison of the different order calculations is performed.
In contrast to the Abelian case, here the g 4, λ-potential suggests a stronger first order phase

2
transition than the g 3 , λ3/2 -potential. Improving the calculation from order g 4 , λ to g 4 , λ2 a
stronger first order phase transition is obtained for both the Abelian and the non-Abelian
case. Of course, this effect questions the reliability of the perturbative approach. The
increase of the surface tension is traced back to the infrared features of typical non-Abelian
diagrams. The observed numerical importance of the λ-corrections has its roots in the
infrared region as well.
The complete standard model results are discussed in sect. 3. The large top quark mass
leads to a decrease of the surface tension. Nevertheless, the qualitative picture is the same
as for the pure SU(2) case.
After some conclusions in sect. 4 the complete analytic result for the standard model is
presented in the appendix.

2 Calculation of the effective potential at finite tem-

perature
2.1 General idea
The effective potential V is calculated using Dyson-Schwinger equations, as described for
the Abelian Higgs model in [12]. A similar way of summing the different contributions to V
for the Φ4 theory has been considered in [14].
Consider a general Lagrangian with interaction terms generating 3- and 4-vertices pro-
portional to g 2 and 3-vertices proportional to g kµ . A generic coupling constant g is used
as an expansion parameter and kµ is a momentum variable. Note that this structure is
suggested by the standard model Lagrangian where the square root of the scalar coupling
√
λ, the Yukawa coupling gY , the electroweak gauge couplings g1 , g2 and the strong gauge
coupling gs play the role of the generic coupling g. Here we give all contributions to the
finite temperature effective potential up to order g 4 . All calculations are carried out in the
imaginary time formalism.
Using the well known technique of Dyson-Schwinger equations the following relation can
be obtained for the effective potential

. (1)

Here the internal lines represent all particles of the theory and ϕ is the “shift” of the La-
grangian in the scalar sector. The two different sorts of blobs are full propagator and full

3
3-vertex respectively. The first term gives
dk
Z
A = tr W(ϕ) . (2)
P
k2 + m2tree + Π(k)
In general, mass, self energy and vertex W are matrices. “tr” denotes the sum over the
suppressed indices. For vector particles the self energies are different for the longitudinal
and transverse part. They can be calculated using the corresponding projection operators
(see [15, 16]). The ϕ-dependence of the propagators is obvious.
The Dyson-Schwinger equation for Π(k), to the order needed in this calculation, reads

. (3)

In the following the indices 2 and 3 denote the contributions of order g 2 and g 3 respectively.
The tadpole part of the self energy can be written as

Πa (k) = Πa2 + Πa2 (k) + Πa3 + · · · with Πa2 (0) = 0 . (4)

In the standard model the only nonvanishing Πa2 (k)-contribution is the longitudinal self
energy of a non-Abelian gauge boson. It is introduced by the corresponding projection
operator when applied to the four vector vertex. The momentum dependence of the third
order term disappears if k0 = 0, thus in the order we are calculating it can be neglected.
The other part of the self energy, Πb (k), has no contribution of order g 2 for scalars.
Nevertheless, for gauge particles those terms do appear. The leading order momentum
independent part of Πb (k) will be called Πb2 .
Using these definitions and introducing the corrected mass term m2 ,

m2 = m2tree + Πa2 + Πb2 , (5)

equation (2) can be written as

Z
dk
A = tr W(ϕ)
P
+k2 m2
+ Πa2 (k) + Πa3 + Πb (k) − Πb2
Z
1 1 1
= tr W(ϕ′ ) dk 2 − 2 Πa3 2
P
2
k + m + Πa2 (k) k + m 2 k + m2
!
1 1 1 1
+ 2 2
Πb2 2 2
− 2 2
Πb (k) 2 . (6)
k +m k +m k +m k + m2

Here the second equality is obtained by expanding the integrand in g. This is best seen by
considering the k0 = 0 and k0 6= 0 parts separately.

4
 Observe that in term B of (1) the vertex need not be corrected to obtain the full g 4 -result.
Inspection of the last term of (6) and term B of (1) shows that their sum is equal to the
derivative −∂V⊖ /∂ϕ, where −V⊖ represents the sum of all two-loop diagrams of the type
shown in fig. 1.a (setting sun diagrams).
With the definitions
!
ϕ 1 1 1
Z Z
′ ′
VR = − dϕ tr W(ϕ ) dk 2 + Π , (7)
P
b2
k + m2 + Πa2 (k) k 2 + m2 k 2 + m2
1 1
Z ϕ Z
Vz = dϕ′ tr W(ϕ′ ) dk 2 Π , (8)
P
a3
k + m2 k 2 + m2
the potential can be given in the form

V = Vtree + V⊖ + Vz + VR . (9)

Note that Vz can be identified as the sum of all terms bilinear in masses coming from two-loop
diagrams of the type shown in fig. 1.b.
Denoting by V3 the sum of the tree level potential and the g 3-order part of VR and calling
V4 the fourth order corrections of VR ,

V3 + V4 = Vtree + VR , (10)

the following final formula is obtained :

V = V3 + V4 + V⊖ + Vz . (11)

It is worthwhile to mention the differences between the method given here and the one
presented in [10]. One advantage of our approach is the absence of thermal counterterms.
The other one is the fact that no different treatment of zero and nonzero Matsubara frequency
modes is needed. Nevertheless, performing the rather tedious calculation using both methods
the above mentioned advantages turned out to be marginal.

2.2 Standard model calculation

To fix our notation the essential parts of the Lagrangian are given

L = LHi}}∫ + L}⊣⊓}⌉ + L{⌉∇mi≀\ + LY⊓k⊣⊒⊣. (12)

Defining the covariant derivative as

Y τa
Dµ = ∂µ + ig1 Bµ + ig2 Wµa (13)
2 2

5
the fermionic and gauge parts are unambiguous. The Higgs contribution reads
!
∞ ϕ3 + iϕ4
LHi}}∫ = −|Dµ ⊕|∈ + ν|⊕|∈ − λ|⊕|△ , where ⊕ = √ (14)
∈ ϕ + ϕ1 + iϕ2

denotes the Higgs doublet. All fermions except the top quark are considered to be massless.
The resulting Yukawa Lagrangian reads
!
tL
LY⊓k⊣⊒⊣ = −}Y 
∐L ♥
⊕⊔R , ∐L = , ♥
⊕ = iτ∈ ⊕∗ . (15)
bL

The calculation is performed in Landau gauge. To define the potential to the order g 4 , λ2
the formal power counting rule
g1 ∼ g2 ∼ gY ∼ λ1/2 (16)

is used. We assume ν = λv 2 to be of order λ, where v is the zero temperature vacuum expec-

tation value of the scalar field. This expansion in the couplings seems to be natural, since it
corresponds to the general structure of the theory, described in the previous subsection. All
masses are treated symmetrically as terms of order g. To the given order the full dependence
on temperature, order parameter ϕ and v is kept.
This clarifies the way, how the general considerations of the previous subsection have
to be applied to the standard model. The different contributions to V⊖ are shown in
fig. 2. As usual, solid, dashed and wavy lines represent fermion, scalar and vector propaga-
tors respectively. In order to make the explicit comparison with the results of [10] easier we
follow the labeling of the setting sun diagrams given there

V⊖ = Va + Vb + Vi + Vj + Vm + Vp . (17)

We have included ghost contributions in Vm . Vp is the scalar setting sun contribution not
considered in the standard model calculation of [10].
The calculation needed for the temperature dependent masses to order g 3 , λ3/2 is similar
to that performed in [9, 16]. With the help of these masses one can evaluate V4 and Vz in
eq. (11). Notice, that working in dimensional regularization the leading order ǫ-dependence
has to be kept in the plasma masses, because it gives finite contributions to the order
g 4 , λ2 due to one-loop divergences. The scalar integral for V⊖ can be found in [17]. More
complicated diagrams of this type can be reduced to the scalar case as described in [10].
These calculations have to be extended to include all contributions of order g 4 , λ2 . After a
long but straightforward calculation the explicit formula for the potential in MS-scheme is
obtained. This final result is given in the appendix. Dropping the appropriate terms of V

6
the lower order g 4 , λ-result, as it is given by Arnold and Espinosa in [10], can be derived.
We have found some minor discrepancies. A careful check of the differences has shown that
some misprints2 in [10] have to be corrected to obtain complete agreement.
We have checked our full g 4 , λ2 -result using the method of [10]. Zero Matsubara fre-
quency modes have been resummed and the necessary temperature counterterms have been
calculated. The obtained potential is in complete agreement with the one we give in the
appendix.
Note, that there are linear ϕ-terms of fourth order in the couplings present in Va and Vz .
These terms cancel each other, thus ensuring the relation limϕ→0 ∂V /∂ϕ = 0 for all allowed
temperatures. This cancellation is essentially the same effect which leads to a vanishing
third order transverse gauge boson mass in the symmetric phase [9], as it can be seen in the
contributions of diagrams fig. 6.o and fig. 6.t of [9].
The result of the present paper with the wave function correction term of [18] gives the
finite temperature effective action up to order g 4 , λ2 .

2.3 Renormalization in the standard model

In case of the standard model it is not possible to avoid the zero temperature renormalization
just by setting µ̄ = 1/β. The reason for that is the large negative gY4 ϕ4 -term, which dominates
over the tree level quartic term. This leads to an MS-potential unbounded from below for
moderately large top mass and small Higgs mass.
We perform a zero temperature renormalization in the on-shell scheme, as described in
[19]. Higgs mass, top quark mass, W- and Z-boson masses and the fine structure constant α
are chosen as physical parameters [20]. The physical masses are the poles of the propagators
and α is defined in the Thompson limit. A multiplicative renormalization of the coupling
constants, the tree level Higgs mass square −ν and the physical Higgs field is performed.
The wave function renormalization of the Higgs field is defined as usual by
∂
δZϕ = Re Πϕ (q 2 ) 2 2 . (18)
∂q 2 q =mϕ,phys

No wave function renormalization is needed for the other fields, because they do not appear
in the effective potential. v is defined to be the true vacuum expectation value of the physical
Higgs field. Therefore it needs no corrections and no tadpole diagrams have to be considered.
The correction to the electric charge δe is gauge independent [21], as it can be easily
checked explicitly using the results of [22]. Therefore in the present calculation the formula
2
σ in eq.(8.2,8.3), eq.(A11) line 9, eq.(A19l), eq.(A19m) line 5, eq.(A19o) line 2, eq.(A25) line 2, eq.(A45)

7
for δe from [19] is used. The logarithmic terms with the five light quark masses are treated
in the way described in [23], with data from [24], resulting in the vacuum polarization
contribution:
γ(5)
Re Πhad (MZ2 ) = −0.0282 ± 0.0009 . (19)

The dependence of the one-loop self energy corrections on the gauge parameters has been
calculated in [22] for gauge bosons. Therefore the corrections in Landau gauge, needed here,
can be taken from [22, 25]. The self energy corrections for the physical Higgs boson and the
top quark can be easily calculated in Landau gauge. Using these quantities the complete
zero temperature renormalization of the potential can be done. The result is thereby freed
of any dependence on µ̄.
Clearly, the analytic expression of these corrections to the potential is too long to be
given here. However, it seems worthwhile to give the numerically most important parts of
the corrections, to enable a simplified usage of the analytic result in the appendix. As it has
already been mentioned, the main contributions come from the gY4 -corrections to parameters
of order λ (see also [10]):

3gY4 mt 3gY4 v 2
δλ = 2
ln , δν = . (20)
8π µ̄ 16π 2
Introducing this corrections in all terms in the potential contributing to order λ and using
standard model tree level relations to calculate the couplings one obtains a result which
is “partially renormalized at zero temperature”. The corresponding correction to the MS-
potential reads
ϕ2
!
1 δλ
δV = −δν + 2 δλ + ϕ4 . (21)
2 2β 4
As we will see it later (sect.4), the numerical effect of this simplification is not too severe.

3 Results for pure SU(2)-Higgs model

3.1 Effective potential and surface tension
To obtain an understanding of the qualitative effects of higher order corrections we study first
the pure SU(2)-Higgs model. In this section the additional U(1)-symmetry and the effect of
the fermions are neglected. A discussion of this simplified version may also be useful in view
of lattice investigations, which will probably deal with the pure SU(2)-Higgs model in the
near future.

8
 The relevant potential can be easily derived from the formulas given in the appendix by
performing the limit g1 , gY −→ 0 and setting the number of families nf to zero. Throughout
this section standard model values for W-mass and vacuum expectation value v are used,
unless stated otherwise : mW = 80.22 GeV and v = 251.78 GeV. The parameter µ̄ of dimen-
sional regularization is set to T = 1/β. This can be justified by the small dependence on
the renormalization procedure. The differences between the results obtained in this scheme
and in a scheme with on-shell T = 0 renormalization are very small. This phenomenon has
been observed in the Abelian Higgs model as well [12].
In fig. 3 different approximations of the effective potential at their respective critical
temperatures are shown. The potentials to order g 3 , λ3/2 and g 4, λ can be obtained from
[8, 9] and [10] respectively. Each approximation suggests a first order phase transition. On
the one hand the critical temperature and the position of the degenerate minimum seems
to be quite stable, on the other hand the hight of the barrier is ∼ 10 times larger for the
g 4 , λ2 case than for the g 3 , λ3/2 -potential. No convergence of the perturbation series can be
claimed for the given parameters.
A more detailed picture can be obtained by considering the surface tension [26]
Z ϕ+ q
σ= dϕ 2V (ϕ, Tc ), (22)
0

which may be seen as a measure of the strength of the phase transition. It can be used
conveniently to discuss the properties of the potential as a function of the Higgs mass. The
results are shown in fig. 4. For very small Higgs masses the third order potential gives a
much larger value for the surface tension than the more complete calculations. The reason
for that is the g 4 ϕ4 contribution, which takes over the role of the tree level λϕ4 term for
small scalar coupling. This radiatively induced quartic term ensures that σ does not increase
for small Higgs masses, a maximum is found. As could have been expected, corrections of
higher order in λ do not change the g 4 , λ result if the scalar mass is small.
This picture changes drastically if larger Higgs masses are considered. In this region
higher order corrections produce an enormous increase in the surface tension. The difference
between the g 4 , λ2 and the g 4 , λ results looks very much the same as in the case of the
Abelian Higgs model (see the discussion in [12]). However, in contrast to the situation there
in the SU(2)-model both curves suggest much larger values of the surface tension than the
g 3 , λ3/2 result.
Let us compare first the results of order g 3 , λ3/2 and g 4 , λ2 . The increase in the strength
of the phase transition, studied already in [11], can be traced back to the infrared features

9
of a non-Abelian gauge theory. The crucial contribution is the one coming from the non-
Abelian setting sun diagram (fig. 2.m). It produces contributions to the potential of type
ϕ2 ln(βmW ) with negative sign. The huge effect of the ln βmW -contribution to the coefficient
of g 4ϕ2 can be understood by recalling that at the critical temperature the leading order
ϕ2 -terms essentially cancel. However, the ϕ2 ln ϕ-type behaviour can not be absorbed in a
correction of Tc , these terms increase the strength of the phase transition. The effect becomes
clear if one deletes the ϕ2 ln βmW -term of Vm by hand. The corresponding surface tension
is shown in fig. 5 (long-dashed line).
We compare now the g 4 , λ [10] and the g 4 , λ2 results. The complete calculation produces
contributions of type ϕ2 ln β(mW + m1,2 ) with positive sign (see appendix). These terms are
coming from the scalar-vector setting sun diagrams (fig. 2.a,b). In [10] scalar masses have
been neglected, resulting in spurious ϕ2 ln βmW -terms with positive sign, which reduces the
surface tension. Another important contribution is the one proportional to g 2 (m1 +3m2 )mW L
from Vz . This term comes from scalar-vector diagrams of type of fig. 1.b and it was neglected
in [10]. On the relevant scale (ϕ < T ) it produces a very steep behaviour of the potential,
again increasing the surface tension. The observed difference between the result of [10] and
the complete g 4, λ2 calculation presented here is mostly due to these two effects, together
with the well known influence of the cubic scalar mass contributions from V3 .
Another interesting effect of higher order λ-corrections is the complete breakdown of
the phase transition at a Higgs mass of about 100 GeV, where the surface tension is very
large. In this region the above mentioned term, proportional to g 2 (m1 + 3m2 )mW L , becomes
important. For a temperature close to the uncorrected barrier temperature Tb , at which the
scalar masses vanish for ϕ = 0, it produces an almost linear behaviour in the small ϕ region.
This results in a potential for which at T = Tb the asymmetric minimum is not a global
minimum but only a local one. Note that Tb is the lowest temperature accessible in this
calculation. In other words, the temperature region in which the phase transition occurs can
not be described by the given method, due to infrared problems.
In order to illustrate the possible effects of the unknown infrared behaviour of the trans-
verse vector propagator, the dependence of the surface tension on the magnetic mass can be
studied. We follow the approach of [9], where a magnetic mass motivated by the solution of
the gap equations was introduced. The transverse vector mass takes the form
!2
γg 2
2
gϕ

m2W = + , (23)
2 3πβ
where γ is some unknown parameter. One can introduce this redefined transverse mass in

10
the most influential infrared contributions, i.e. in the m3W - and in the ϕ2 ln βmW -terms. We
show in fig. 5 the results obtained for γ = 0, 2 and 4. The qualitative behaviour is similar
to results found in [9]. The main difference is due to the fact that the higher order result
suggests a stronger first order phase transition, thus for a given mHiggs a larger magnetic
mass is necessary to change the phase transition to second order.
A complete fourth order calculation of the surface tension has to include the wave function
correction term Zϕ (ϕ2 , T ) calculated in [18]. Using the results of [18] we have determined
σ for Higgs masses between 25-95 GeV. The numerical effect of this Z-factor is very small,
only 1% − 4%.

3.2 Further properties of the potential

The latent heat of the phase transition is another interesting quantity to be calculated from
the effective potential :
∂
∆Q = T V (ϕ+ , T ) , (24)
∂T Tc

where ϕ+ is the position of the asymmetric minimum of V . We have plotted ∆Q as a function

of mHiggs in fig. 6. The latent heat of the higher order calculations (g 4 , λ and g 4 , λ2 ) increases
almost linearly with the Higgs mass. This somewhat surprising behaviour can be understood
by observing that for those potentials neither the position of the degenerate minimum nor
the height of the barrier change significantly with increasing Higgs mass (see fig. 4). On the
other hand the critical temperature is essentially proportional to mHiggs .
For completeness, the quantity ϕ+ /Tc , relevant for baryogenesis, is shown in fig. 7 as a
function of the Higgs mass. It is interesting to observe that the upper part of the region
favouring baryogenesis [2], i.e. ϕ+ /Tc ≈ 1 at mHiggs ≈ 40 GeV, coincides with the region
of best reliability of the perturbative approach. As has already been pointed out in [5, 12],
this parameter does not reflect the dramatic change of the potential at critical temperature
introduced by higher order corrections.
Now the question arises whether a good convergence of the perturbation series, which
can not be claimed in the whole range of λ for a realistic gauge coupling g = 0.64, could
be present in the region of much smaller gauge coupling constants. This seems indeed to be
the case, as can be seen in fig. 8, where the surface tensions of order g 3 , λ3/2 and g 4 , λ2 are
plotted for a model with a vector mass of 20 GeV, i.e. g = 0.16. In the used Higgs mass
range the two results for σ differ by a factor of two at most. The relative size of this range,
i.e. the ratio of the minimal and maximal values of the Higgs mass, is 4, which is twice as

11
large as the range for the model with mW = 80 GeV.

4 Standard model results

In the case of the full standard model the qualitative behaviour of the potential is essen-
tially the same as for the SU(2)-Higgs model. The main difference is a decrease of the
surface tension by a factor ∼ 4. This can be traced back to the large top mass. Also
the characteristic points of the surface tension plot of fig. 4 are shifted to higher values of
the Higgs mass. We show σ as a function of mHiggs in fig. 9. The complete breakdown
of the g 4, λ2 calculation, observed at mHiggs ≈ 100 GeV for the pure SU(2) case, occurs
at mHiggs ≈ 200 GeV in the full model. These quantitative differences do not change the
qualitative features of the potential, thus the discussion given in the previous section does
also apply to the standard model. The difference between the fully renormalized potential
and the partially renormalized potential (see eq. (20),(21)) is not too severe in view of the
huge uncertainties still present in the perturbative approach. Again, the position of the sec-
ond minimum at the critical temperature, given in fig. 10, does not depend as strongly
on the order of the calculation as the height of the barrier. Unfortunately, the region
mHiggs ≈ 40 GeV, in which the reliability of the perturbative approach is the best and
ϕ+ /Tc ≈ 1, is well below the experimental Higgs mass bound.

5 Conclusions
In the previous sections we have calculated and analyzed the finite temperature effective
potential of the standard model up to order g 4 , λ2 . We have determined several physical
quantities as functions of the Higgs mass. However, to the given order the systematic ex-
pansion in coupling constants does not permit a definitive statement about the character of
the phase transition for realistic Higgs masses. This is seen from the fact that the g 4 , λ2 -
corrections are huge and even the step from a g 4 , λ-calculation to the complete g 4 , λ2 -result
changes the potential essentially if the Higgs mass is large. One source of the dramatic
increase of the surface tension are the infrared contributions of the typical non-Abelian di-
agrams. Quantitative information on a possible infrared cutoff, e.g. a magnetic mass term,
could increase the reliability of the calculation drastically. For large Higgs masses another
infrared problem is connected with the scalar sector, namely the corrected leading order
scalar masses vanish near the critical temperature producing an almost linear term in the

12
potential. It has to be concluded, that although resummation techniques permit a system-
atic expansion in coupling constants, the numerical results still point to the unknown low
momentum behaviour of the theory as the main obstacle of any reliable prediction.

Special thanks go to W. Buchmüller for his continuous support of this work. Helpful discus-
sions with D. Bödeker, T. Helbig, B. Kniehl and H. Kohrs are also acknowledged. Z. F. was
partially supported by Hung. Sci. Grant under Contract No. OTKA-F1041/3-2190.

Appendix
Here the different contributions to the potential described in sect. 2 are given explicitly. The
formulas have been simplified as much as possible to enable a direct numerical use and further
analytic investigation. The authors3 are ready to supply a FORTRAN code evaluating the
different approximations of the effective potential (g 3, λ3/2 ; g 4, λ; g 4 , λ2 pure SU(2) and
standard model, with complete on-shell renormalization or partial renormalization) as a
function of ϕ and T .
Linear mass terms, poles in 2ǫ = 4 − n and terms proportional to the constant ιǫ (see
[10]), which cancel systematically in the final result, are not shown and the limit n → 4 has
already been performed. The leading order resummed scalar masses are given by
1 9 3
 
m21 = 2λϕ + 2
m22 , m22 2
= λϕ − ν + 2
6λ + g22 + g12 + 3gY2 , (25)
12β 4 4
while the transverse vector boson masses and the fermion mass remain uncorrected to leading
order :
1 1
mW = g2 ϕ , mf = √ gY ϕ .
mZ = mW / cos θW ,
(26)
2 2
The longitudinal SU(2)×U(1) mass matrix receives temperature corrections in the diagonal
elements [8]

1 g2 5 1 1 g2 1 5
   
m2W L = g22ϕ2 + 22 + nf , m2BL = g12 ϕ2 + 12 + nf , (27)
4 β 6 3 4 β 6 9
which result in longitudinal masses defined by

m2W L − 14 g1 g2 ϕ2 m2ZL
     
cos θ̃ sin θ̃ 0 cos θ̃ − sin θ̃
= . (28)
     
   
− 41 g1 g2 ϕ2 m2BL − sin θ̃ cos θ̃ 0 m2γL sin θ̃ cos θ̃
3
e-mail fodor@vxdesy.desy.de or t00heb@dhhdesy3.bitnet

13
In the following the short hand notations

s = sin θW , c = cos θW , s̃ = sin θ̃ , c̃ = cos θ̃ (29)

are used. Evaluating the scalar one- and two-loop temperature integrals the constants

3
c0 = 2
+ 2 ln 4π − 2γ ≈ 5.4076 and c2 ≈ 3.3025 (30)

are introduced following [27] and [17] respectively. Now all the contributions to the potential,
which have to be summed according to formulas (11) and (17), can be given explicitly :

ϕ2
" #
1 1 3 1 1 λ

V3 = −ν + 2 λ + g22 + g12 + gY2 + ϕ4 (31)
2 β 2 16 16 4 4

1 h 3 i
− m1 + 3 m2 3 + 4 mW 3 + 2 mW L 3 + 2 mZ 3 + mZL 3 + mγL 3
12πβ

g22
" !
1 1 β 1 1 1 1
 
2
Va = mW 2 − + ln − ln µ̄2 β 2 − c0 + c2 + (32)
32π 2 β 2 c2 2c4 3 12 6 4 4
 ( ! )
1 1 β
 
+ 1+ 2 m21 + 3m22 −4 ln + ln µ̄2 β 2 − c2 + 2m2 (m1 + m2 )
4 2c 3

1 1
 
−4s2 m22 ln (2m2 ) − 1+ (m1 − m2 )2 (m1 + m2 ) + 2m2 mZ s2
2mW 2c

mW 1 3  2
 
2 2

− 1 + 3 (m1 + 3m2 ) − m1 − m2 ln(m1 + m2 )
2 2c 4m2W

1
    2 
+ m4W − 2 m21 + m22 m2W + m21 − m22 ln (m1 + m2 + mW )
2m2W

1
    2 
+ 2 m4Z −2 m21 + m22 m2Z + m21 − m22 ln (m1 + m2 + mZ )
4mW
#
1 1 2
   
2 2 2 2
+ − s mZ − 4m2 ln(2m2 + mZ ) + mW − 4m2 ln(2m2 + mW )
4c2 2

g22 s4 m4
" ( )
1 4 1
  
Vb = c + 4 + 4c2 + 2 − 10 m2W + 2 c2 − 2 s2 m22 + 2 2
4
(33)
64π 2 β 2 c c c mW

14
 ( )

2
 1  2
× ln(mW + mZ + m2 ) + 5 − 4c m2W − 2 m2W c2 + m22 s2 ln(mW + m2 )
mW

s4  s4
( ! )
1 1
2 
− 2 m2Z − m22 ln(mZ + m2 ) + mW m2 2 − c2 + + m1 1+ 3
mW c c 2c

c2 5 s2 2
( ! )
5 5 5 β 2 1
 
+m2W + − 2 ln + c2 + − + 2 + c − 2
2c2 8c4 4 9µ̄ 2 2 c c c

1
  2  2 
− 2 2 m2W − m21 ln(mW + m1 ) + m2Z − m21 ln(mZ + m1 )
2mW

m41 m41
! !
1
+ 4m2W − 2m21 + ln(2mW + m1 ) + 2 2m2Z − m21 + ln(2mZ + m1 )
2m2W c 4m2Z

s2 m2
!
1 1 3
   
+s 2
m22 2 ln m2 + + 1 1+ 2 + 2 s 4
m42 ln m2 + m41 ln m1
c 2 2c mW 4

ϕ2 n
+ (g2 c̃ + g1 s̃)4 ln(2mZL + m1 ) + (g2 s̃ − g1 c̃)4 ln(2mγL + m1 )
4g22

+2g24 ln(2mW L + m1 ) + 2(g2 c̃ + g1 s̃)2 (g2 s̃ − g1 c̃)2 ln(mZL + mγL + m1 )

#
 o
+4g22g12 2
s̃ ln(mW L + mZL + m2 ) + c̃ ln(mW L + mγL + m2 ) 2

g22 m2f  g22 nf m2W 10

"( )
1 17 5
  
2 2
Vi = 10 + 2 + − 4 − 14 + gs mf (34)
8π 2 β 2 96 c 36 c2 c

4g 2 nf m2W
#
1 10 10 5
 
×(ln µ̄2 β 2 − c0 + + ln 2) − 2 2
− 4 − 14 ln 2
2 3 27 c c

gY2 3
   
2 2 2 2 2 2
Vj = 9mf − m1 − 3m2 ln µ̄ β − c0 + − ln 4 + 48mf ln 2 (35)
128π 2 β 2 2

g22 c4 
"
2
 1 1 5 2
Vm = m − − + − c − ln(mW + mZ ) (36)
16π 2 β 2 W 4c4 c2 2 4

15
 1 1 31 11 51 251
 
+ 4
+ 2 − 5 − 4c2 ln(2mW + mZ ) + ln µ̄2 β 2 − c0 − c2 −
8c c 8 16 16 96

11 5 1 1 1 1 1 51 β
 
− c− + 2 − 4s2 ln 2 + c2 (c − ) + 2 − 4 ln c − ln
12 4c 8c 4 2 8 c 4 3
!
1 23 1 4
 
2
+ − + 5c + c ln mW − mW mW L (1 + c) − 2s2 m2W L ln(2mW L )
8c4 4 4

1 1 m3
 
+ m2W − 2c2 m2W L ln(2mW L + mZ ) + m2W L − W L
2 2 mW
  2 
m2W L − m2γL
(
+s̃2 m2W − 2m2W L − 2m2γL +  ln(mW L + mγL + mW )
 
m2W
 2
m2W L mγL + mW L m2γL − m3γL m2W L − m2γL ln(mγL + mW L )
)
+mγL (mW L − mW ) + −
mW m2W
( )#
2
+c̃ mγL −→ mZL

3λ2 ϕ2 9µ̄2
" #
Vp = − ln − c2 − 2 ln{m1 (m1 + 2m2 )} (37)
32π 2 β 2 β2

ϕ2 g24
"
1 1 35 293 1 13
  
2 2
V4 = − − (ln µ̄ β − c 0 ) + − − (38)
64π 2 β 2 4 c2 2c4 4 72 18c2 18c4

g 4 nf g22s2 3gY2 2
! !
10 5 2 gY 1
  
+ 2 2
− 4 − 14 + gY2 − + 4g 2
s ln 4 + g 2 +λ 1+ 2
27 c c 3c2 4 2 2c
#
3 1 n

+3gY2 λ 2 2
ln µ̄ β − c0 + − − 4m4W − 2m4Z − 24m4f ln 4
2 64π 2

 3 o
 
+ 6m4W + 3m4Z + m41 + 3m42 − 12m4f 2 2
ln µ̄ β − c0 +
2

16
 "
1 n  
Vz = 1
4
(m1 + 3m2 ) g22 2mW L + 4mW + mZL c̃2 + 2mZ c2 (39)
32π 2 β 2

   o
+g12 mZL s̃2 + 2mZ s2 + mγL g22 s̃2 + g12 c̃2

 
+ 12 g2 g1 (m1 − m2 ) {2mZ sc + (mZL − mγL )s̃c̃} + 4g22 mW L mW + mZ c2

#
   
+g22 mW 2
4mγL s̃ + 38 mW + 16
3
mZ c 2
+ 4mZL c̃ 2
+ 3λ m1 m2 + 1 2
m
2 1
+ 5 2
m
2 2
.

References
[1] V. A. Kuzmin, V. A. Rubakov and M. E. Shaposhnikov, Phys. Lett. B155 (1985) 36

[2] A. G. Cohen, D. B. Kaplan and A. E. Nelson, Annu. Rev. Nucl. Part. Sci. 43 (1993) 27

[3] A. Jakovác and A. Patkós, Z. Phys. C60 (1993) 361;

A. Jakovác, K. Kajantie and A. Patkós, preprint HU-TFT-94-01

[4] P. Ginsparg, Nucl. Phys. B170 [FS1] (1980) 388;

J. March-Russel, Phys. Lett. B296 (1992) 364;
M. Gleiser and E. W. Kolb, Phys. Rev. D48 (1993) 1560

[5] P. Arnold and L. G. Yaffe, preprint UW/PT-93-24 (1993)

[6] M. Reuter and C. Wetterich, Nucl. Phys. B408 (1993) 91;

M. Reuter and C. Wetterich, preprint DESY-94-017 and HD-THEP-93-41

[7] B. Bunk, E.-M. Ilgenfritz, J. Kripfganz and A. Schiller, Nucl. Phys. B403 (1993) 453;
K. Kajantie, K. Rummukainen and M. Shaposhnikov, Nucl. Phys. B407 (1993) 356

[8] M. E. Carrington, Phys. Rev. D45 (1992) 2933

[9] W. Buchmüller, Z. Fodor, T. Helbig and D. Walliser, preprint DESY-93-021 (1993),

Ann. Phys., to appear

[10] P. Arnold and O. Espinosa, Phys. Rev. D47 (1993) 3546

[11] J. E. Bagnasco and M. Dine, Phys. Lett. B303 (1993) 308

17
[12] A. Hebecker, Z. Phys. C60 (1993) 271

[13] S. Weinberg, Phys. Rev. D9 (1974) 3357;

D. A. Kirzhnits and A. D. Linde, JETP 40 (1974) 628

[14] C.G. Boyd, D.E. Brahm and S.D.H. Hsu, Phys. Rev. D48 (1993) 4963

[15] J.I. Kapusta, Finite temperature field theory (Cambridge Univ. Press, 1989) pp. 69–72

[16] W. Buchmüller, T. Helbig and D. Walliser, Nucl. Phys. B407 (1993) 387

[17] R.R. Parwani, Phys. Rev. D45 (1992) 4695

[18] D. Bödeker, W. Buchmüller, Z. Fodor and T. Helbig, preprint DESY-93-147 (1993),

Nucl. Phys., to appear

[19] M. Böhm, H. Spiesberger and W. Hollik, Fortschr. Phys. 34 (1986) 687

[20] Particle Data Group, K. Hikasa et al., Phys. Rev. D45 (1992) S1

[21] A. Sirlin, Phys. Rev. D22 (1980) 971

[22] G. Degrassi and A. Sirlin, Nucl. Phys. B383 (1992) 73

[23] W. F. L. Hollik, Fortschr. Phys. 38 (1990) 165

[24] F. Jegerlehner, in Testing the Standard Model—Proceedings of the 1990 Theoreti-

cal Advanced Study Institute in Elementary Particle Physics, ed. by M. Cvetič and
P. Langacker (World Scientific, Singapore, 1991) p. 476

[25] W.J. Marciano and A. Sirlin, Phys. Rev. D22 (1980) 2695

[26] S. Coleman, Phys. Rev. D15 (1977) 2929;

A. D. Linde, Nucl. Phys. B216 (1983) 421

[27] L. Dolan and R. Jackiw, Phys. Rev. D9 (1974) 3320

18