Pure Appl. Chem.

2015; 87(9-10): 937–951

Conference paper

Bernd Herzog* and Uli Osterwalder

Simulation of sunscreen performance

DOI 10.1515/pac-2015-0401

Abstract: Sunscreens are used to protect the human skin against harmful effects of solar UV radiation. The
most important quantity characterizing sunscreen performance is the sun protection factor (SPF). At the
stage of development of new sun protection formulations quick and inexpensive methods for estimation of
the UV screening performance are highly desirable. The most convenient approach towards this goal is given
by computational simulations. Models for the calculation of the SPF employ the same algorithm as used with
in vitro SPF measurements, but replace the transmittance measurement by the calculation of the overall
absorbance of the UV filters in an irregular sunscreen film. The simulations require a database with quantita-
tive UV extinction spectra of the relevant UV filters as well as a mathematical description of the film irregular-
ity. The simulation algorithm implies also the consideration of photodegradation properties of the UV filters
in the sunscreen composition. Besides using such simulations for designing new sunscreen formulations, the
calculations can also support the understanding of sunscreen performance in general.

Keywords: computer simulation; irregular film model; kinetics; Photobiology-16; sun protection factor;
­sunscreens; UV-visible spectroscopy.

Introduction
It is a well-known fact that an overexposure of human skin by UV light may lead to sunburn and an increased
risk for skin cancers [1–3]. With our current knowledge, these effects can be attributed to the UVB part (290–
320 nm) as well as to the UVA part (320–400 nm) of the solar spectrum [4]. Accounting for the much higher
abundance of UVA in solar radiation compared to UVB, UVA contributes to the development of erythema to
about 15 % of the corresponding effect of the overall solar UV radiation. The sun protection factor (SPF) is a
measure of how much the erythemally effective radiation is reduced by a sunscreen when applied on human
skin. It is defined as the ratio of the minimal erythemal doses (MED) of simulated solar radiation directed
to human skin in the presence and in the absence of a sunscreen agent [5]. Based on this definition, in vivo
methods for sunscreen testing on volunteers have been established [6–9]. At the time being, such in vivo
methods have to be applied before claiming SPF values on sunscreen products.
As testing on volunteers is time-consuming and expensive, for purposes of experimental screening in
vitro methods for determination of the SPF have been introduced [10–12]. Such methods are based on meas-
urements of UV transmittance of sunscreen films. Substrates on which the sunscreen films are applied model
in some way the inhomogeneous surface structure of the human skin, such as polymethylmethacrylate

Article note: A collection of invited papers based on presentations at the 16th International Congress on Photobiology (ICP-16),
Córdoba, Argentina, 7–12 September 2014.

*Corresponding author: Bernd Herzog, BASF Grenzach GmbH, Koechlinstrasse 1, 79639 Grenzach-Wyhlen, Germany,
e-mail: bernd.herzog@basf.com
Uli Osterwalder: BASF Personal Care and Nutrition GmbH, 40789 Monheim, Germany

© 2015 IUPAC & De Gruyter

938 B. Herzog and U. Osterwalder: Simulation of sunscreen performance

(PMMA)-plates with rough surface [12]. Measurements of UV transmittance are based on the assumption that
the UV-protection of sunscreens is only caused by the attenuation of UV-light. Any other effects, which might
be of relevance in vivo, like anti-inflammation or anti-oxidative properties, are not considered.
The sun protection factor is not unequivocally related to the overall UV absorbance spectrum of a sun-
screen (absorbance, extinction, and optical density are synonymous terms). That means, it is possible to have
quite different levels of UVA protection with the same SPF value. For that reason the protection against UVA
has to be assessed with an independent method specific for this spectral range. The most frequently used in
vivo method, persistent pigment darkening (PPD), based on the pigmentation of volunteers after irradiation
with a pure UVA light source, has been widely taken over by an in vitro method, based on UV transmittance of
a sunscreen film and the determination of an UVA protection factor (UVAPF) by adjusting that measurement
to the corresponding in vivo SPF [13, 14]. There are several other in vitro methods for the assessment of UVA
protection in place [15], all based on measurements of UV transmittance.
Knowing the amounts and characteristics of the UV filter substances of a sunscreen, it is possible to cal-
culate its UV transmittance, when taking also film irregularity and photodegradation kinetics into account.
From simulated UV transmittance, SPF and all quantities characterizing protection against UVA can be
computed.

Experimental
Some experimental data are necessary for the calculations, such as quantitative UV extinction spectra of the
UV absorbers used and the kinetic constants of photodegradation.
For spectroscopic measurements ethanol (p.a., Merck) and bidistilled water were used as solvents. The
­UV-absorbers used in this work are given below. After the chemical name the INCI name is given in parenthesis
­followed by the abbreviation and the trade name. The supplier is also indicated. 3-(4-methoxyphenyl)-­2-ethylhexyl
ester (ethylhexyl methoxycinnamate, EHMC, Uvinul® MC80) from BASF; 2-phenylbenzimidazole-5-sulfonic
acid (2-phenylbenzimidazole-5-sulfonic acid, PBSA, Eusolex® 232) from Merck; 2-hydroxy-4-methoxy benzo­
phenone (benzophenone 3, B3, Uvinul® M40) from BASF; titanium dioxide (TiO2, Eusolex® T2000) from
Merck; trianilino-(p-carbo-2′-ethylhexyl-1′-oxi)-1,3,5-triazine (ethylhexyl triazone, EHT, Uvinul® T150) from
BASF; 4-t-butyl-4′-methoxydibenzoyl methane (butyl methoxydibenzoyl methane, BMDBM, Parsol® 1789)
from DSM; 2-ethylhexyl-2-cyano-3,3-diphenyl-2-propenoate (octocrylene, OCR, Uvinul® N539) from BASF;
2-ethylhexyl p-dimethylaminobenzoate (ethylhexyl dimethyl PABA, OD-PABA, Eusolex® 6007) from Merck;
2,4-bis-{[4-(-2-ethyl-hexyloxi)-2-hydroxy]-phenyl}-6-(4-methoxyphenyl)-(1,3,5)-triazine (bis ethylhexylphe-
nol methoxyphenyl triazine, BEMT, Tinosorb® S) from BASF; 2,2′-Methylene-bis-(6-(2H-benzotriazole-2-yl)-
4-(1,1,3,3-tetramethylbutyl)-phenol (Methylene Bis-Benzotriazolyl Tetramethylbutylphenol, MBBT, Tinosorb®
M) from BASF; 2-(4-diethylamino-2-hydroxybenzoyl)-benzoic acid hexylester (diethylamino hydroxybenzoyl
hexyl benzoate, DHHB, Uvinul® A Plus) from BASF; 2-(2H-benzotriazol-2-yl)-4-methyl-6-[2-methyl-3-[1,3,3,3-
tetramethyl-1-[(trimethylsilyl)oxy]disiloxanyl]propyl]-phenol (dromtrizole trisiloxane, DTS, Mexoryl® XL);
3,3′-(1,4-phenylendimethylene)-bis-(7,7-dimethyl-2-oxo-bicyclo-[2.2.1]heptane-1-methane sulfonic acid (tereph-
thalidene dicampher sulfonic acid, TDSA, Mexoryl® SX). The latter two compounds were synthesized at BASF.
UV-spectroscopic transmission measurements were carried out using a Perkin Elmer Lambda 16 spec-
trometer. UV-transmission spectra of particulate UV-absorbers were recorded with the same instrument using
an integration sphere attachment (Labsphere B009-4012). UV-spectra of oil-soluble filters have been meas-
ured quantitatively in diluted ethanolic solutions in UV transparent cuvettes of 1 cm optical thickness. UV
spectra of the particulate filters MBBT and of TiO2 were measured quantitatively with dispersions containing
the UV-absorber between 0 and 3 % and sandwich-type cuvettes with an optical pathlength of 0.0008 cm
[16]. Photodegradation kinetics of single UV filters and also of binary mixtures were measured in sunscreen
formulations as reported previously [17, 18].
In vivo SPF data of the sunscreen formulations used in this study were determined according to the Inter-
national SPF Test Method [8] or the corresponding ISO standard 24444 [9].
 B. Herzog and U. Osterwalder: Simulation of sunscreen performance 939

Theoretical aspects

Sun protection factors from UV transmittance

The basic principle of sun protection factor calculations is the calculation of the factor by which the intensity
of the UV radiation is reduced due to the presence of a sunscreen. This factor is given by the inverse of the UV
transmittance of the absorbing film, 1/T. At a certain wavelength λ, 1/T(λ) is also designated as monochro-
matic protection factor (MPF). As the spectral range relevant for the formation of erythema is between 290
and 400 nm, the monochromatic protection factors should be averaged over this range. In order to obtain
the SPF, this average must be weighted with the intensity of the light source, Ss(λ) and the erythemal action
spectrum, ser(λ), leading to eq. (1) contributed by Sayre et al. [19]:
400

SPF =
∑ 290
ser ( λ ) ⋅ Ss ( λ )
(1)
400
∑ 290
ser ( λ ) ⋅ Ss ( λ ) ⋅ T ( λ )


Data for Ss(λ) and ser(λ) are available [10, 20], but T(λ) has to be determined for the respective sunscreen
either by in vitro measurement or by calculation. Ss(λ) is given in units of W m−2 nm−1, ser(λ) without units.
The product of Ss(λ) and ser(λ) is called erythemal effectiveness spectrum or erythemally effective irradiance,
Eer(λ), and again has units of W m−2 nm−1. The quantities Ss(λ) and ser(λ) are plotted in Fig. 1. In Fig. 2 the
erythemally effective irradiance in the case of no protection, Eer(λ), is compared to situations of protection,
Eer(λ)·T(λ), with sunscreens of SPF = 10 and SPF = 20.

1 1

0.8 0.8
Ss(λ), (W m–2nm–1)

ser(λ),(without units)

0.6 0.6

0.4 Erythemal action spectrum 0.4

Standard sun irradiance

0.2 0.2

0 0
290 400
Wavelength (nm)

Fig. 1: Erythemal action spectrum and spectrum of standard sun irradiance.

0.015
Erythemal effectiveness

0.01
(W m–² nm–1)

Without sunscreen

SPF = 10

0.005 SPF = 20

0
290 400
Wavelength (nm)

Fig. 2: Erythemal effectiveness in absence and presence of sunscreens.

940 B. Herzog and U. Osterwalder: Simulation of sunscreen performance

Overall UV spectrum of a UV filter composition in a sunscreen

Sunscreens in most cases contain more than one UV filter. In order to obtain the overall absorbance of a UV
filter composition via Beer-Lambert’s law, the overall molar concentration c̅ (from the average molecular
weight M̅) and the average molar extinction coefficient ε̅(λ) are calculated via eq. (2),

n n βi
M = ∑ i = 1 βi / ∑ i = 1 , (2a)
Mi

n
c = 10 ⋅ ∑ i = 1 βi / M , (2b)


n εi ( λ ) ⋅ βi n βi
ε( λ ) = ∑ i = 1 ∑ (2c)
Mi i=1
Mi

where Mi is the molar mass, βi the concentration in percent (w/v), and εi(λ) the molar extinction coefficient of
an individual UV absorber [16, 17].
In order to calculate transmittance via Beer–Lambert’s law, the only quantity still missing is the thick-
ness d of the sunscreen film on the skin:

T ( λ ) = 10 − ε ( λ ) cd (3)

In the in vivo method for SPF determination, 2 mg/cm2 sunscreen are applied on human skin, approxi-
mately corresponding to 2 μL/cm2. This results in a film of 20 μm (0.002 cm) thickness. However, this film is
not homogenous, but irregular. As water evaporates, the film will get thinner, but at the same time the con-
centrations of the UV-filters will increase accordingly, such that the product c·d will stay constant.
The above calculation of the film thickness is based on the volume of the sunscreen spread on certain
skin area. The error made here taking the volume instead of the weight cancels with the error assuming
the filter concentrations given in weight per volume, as in practice formulations are prepared in weight per
weight.

Accounting for film irregularity

When measuring UV transmittance in vitro, in some way the irregular surface structure of the human
skin has to be modeled by using appropriate substrates with rough surface. This is important because the
optical transmittance of an absorbing film of uniform thickness always is lower compared to that of a cor-
responding irregular film of the same average thickness. This fact is illustrated in Fig. 3.

T(λ) = 10–E(λ)

E = 1.0 E = 1.5
E = 0.5
–1.5
T = 10 –1
= 0.1 T = 0.5 · 10 + 0.5 · 10–0.5 = 0.174
1/T = 10 1/T = 5.7

Homogeneous film Irregular film

High SPF
Low SPF

Fig. 3: A homogeneous film always absorbs light better than an irregular film.
 B. Herzog and U. Osterwalder: Simulation of sunscreen performance 941

The effect can be quite dramatic, and several models describing sunscreen film irregularity have been
published [21–24]. The model of Ferrero and coworkers [23], which is employed in the simulations reported
here, uses an approach common in surface metrology. In this case, the film profile becomes equivalent to
the bearing area curve of Abbot and Firestone [25, 26] and is constructed based on a cumulative distribution
function F(h) containing the film height h as a random variable. Among possible probability functions, the
authors chose a gamma law representing an asymmetrical distribution, and f(h) is the associated probability
density function:
c−1  h
 h  − b  1
f ( h) =   e , (4)
 b bΓ( c )

where h is the random variable “relative height,” c is the shape parameter to be adjusted, b is needed for
normalization and Γ(c) is the value of the gamma function at c. The cumulative height distribution F(h) is
obtained by integration of f(h) from 0 to ∞:
∞
F ( h ) = ∫ f ( h ) dh. (5)
 0

In order to construct the film thickness profile, h is deduced from its cumulative distribution F(h), ranging
from 0 to 1. As the area under the film profile is normalized to 1, the following relationship must hold:
1
∫ h( F ) dF = 1 
0
(6)

The transmittance of the film can then be calculated:

1 − ε ( λ ) cdh ( F )
Tfilm ( λ ) = ∫ 10 dF (7)
0

The transmittance calculated via eq. (7) can be fed into eq. (1), resulting in an SPF value. The shape para-
meter c is the screw by which the calculated result can be adjusted to fit the in vivo SPF. In Fig. 4, the density
function given in eq. (4) and its cumulative version indicated in eq. (5) are depicted for a shape parameter of c =
1.5. Figure 5 shows the respective film profile defined in eq. (6). In Fig. 6 the influence of the shape parameter is
demonstrated for a sunscreen composition with in vivo SPF of 15.5. The larger the shape parameter, the higher
the calculated SPF value. Obviously, the value of the SPF is largely determined by the thickness of the thinner
parts of the film. In Fig. 5 the film profile of the irregular film is drawn as well as the profile of the corresponding

0.8

0.6

0.4
f(h)

0.2

0
0 1 2 3 4 5 6
h
1
F(h)

0
0 1 2 3 4 5 6
h

Fig. 4: Probability density distribution f(h) of film thickness h and the corresponding cumulative distribution F(h).
 942 B. Herzog and U. Osterwalder: Simulation of sunscreen performance

4
h

Normalized average
height = 1
0
0 0.2 0.4 0.6 0.8 1
F(h)
Parent film

Fig. 5: Construction of irregular film profile by interpreting F(h) as film extension.

c = 0.5, SPF = 6.3

6
c = 1.0, SPF = 17.3

c = 1.5, SPF = 35.0

4
h

0
0 0.2 0.4 0.6 0.8 1
F(h)

Fig. 6: Structure of irregular film with different values of shape parameter c; corresponding SPF data calculated for the P3
standard sunscreen (composition: 2.78 % PBSA, 3 % EHMC, 0.5 % BMDBM).

2.0
c = 0.5, SPF = 6.3
c = 1.0, SPF = 17.3
1.5 c = 1.5, SPF = 35.0
E (irregular film)

1.0

0.5

0.0
0 2 4 6 8 10
E (parent film)

Fig. 7: Absorbance of irregular film as function of parent film absorbance, calculated for P3 standard sunscreen (composition:
2.78 % PBSA, 3 % EHMC, 0.5 % BMDBM) with different values of shape parameter c.

regular film, also called parent film. When plotting the irregular film extinction (E, synonymous with absorb-
ance) against parent film extinction, a non-linear relationship is observed, shown in Fig. 7 for different shape
parameters and the composition of the P3 standard formulation [6], but without treatment of photoinstabilities.
 B. Herzog and U. Osterwalder: Simulation of sunscreen performance 943

Changes of UV transmittance due to photoinstabilities

Equation (1) refers to a static value of the SPF without considering any changes of UV transmittance due to
photoinstabilities, which may occur in the course of the irradiation process involved with SPF determination
in vivo. In order to take this into account when performing in vitro SPF measurements, eq. (1) was modified
by Stanfield et al. [27], with tMED referring to the time, after which one minimal erythemal dose (1 MED) is
transmitted through the sunscreen film:
t MED 400

SPF =
∑ ∑ 0 290
ser ( λ ) ⋅ Ss ( λ )
(8)
t MED 400
∑ ∑ s
0 290 er
( λ ) ⋅ Ss ( λ ) ⋅ T ( t , λ )

This approach mimics the conditions of SPF determination in vivo. A similar and equivalent method
was described by Wloka et al. [28]. These procedures were developed for in vitro SPF measurements, but can
equally be applied when UV transmittance is calculated, under the prerequisite that decay constants of the
photodegradation of the UV filters are known [17, 18]. The kinetics of UV filter photodegradation is approxi-
mated with a first-order law, where ci is the molar concentration of UV filter i (ci,0 without irradiation, ci,UVdose
after irradiation with a certain UV dose), and ki is its kinetic constant of photodegradation:

ci ,UV dose = ci ,0 ⋅ exp( −ki ⋅ UV dose ) (9)

For practical reasons, instead of time, UV dose is written in eq. (9), which is proportional to irradiation
time. There can be stabilizing or destabilizing interactions between different UV filters. This is accounted for
by formalisms which decrease the ki in case of stabilization and increase them in case of destabilization. For
stabilization the following equation was found [18]:

ki ,j = ( 1/ ki ,0 + q j β j ) −1 , (10),

where ki,0 is the degradation constant without stabilization, qj is the stabilizer constant of stabilizer j, and βj
is its concentration in percent (w/v). For destabilization eq. (11) is employed:
ki ,m = ki ,0 ( 1 + rm βm ), (11)

where ki,0 is the degradation constant without destabilization, rm is the destabilization constant of destabi-
lizer m, and βm is its concentration in percent (w/v). In Table 1 values of ki are given for the UV filters used
in this work. Table 2 lists stabilization and destabilization constants. By determining the interaction factors
ki,j/ki,0 and ki,m/ki,0, any number of interactions can be taken into account by multiplying ki,0 with all existing
interaction factors.

Table 1: Photodegradation constants of UV filters used in this work.

UV filter Photodegradation constant ki (MED−1)

OCR 0.0010
PBSA 0.0047
EHT 0.0058
OD-PABA 0.0139
EHMC 0.0417
B3 0.0010
DTS 0.0010
BEMT 0.0010
MBBT 0.0002
DHHB 0.0021
TDSA 0.0058
BMDBM 0.2200
944 B. Herzog and U. Osterwalder: Simulation of sunscreen performance

Table 2: Stabilizing constants qj and destabilizing constants rm.

UV filter combination (the second Stabilizing constants Destabilizing constants

filter is the stabilizer or destabilizer) qj (MED · %−1) rm (%−1)

BMDBM + OCR 2.33

BMDBM + BEMT 0.90
EHMC + OCR 2.33
EHMC + BEMT 0.90
EHMC + MBBT 0.90
EHMC + BMDBM 0.244
BMDBM + EHMC 0.012

In addition, the photostability of photounstable UV filters can be improved when UV light is strongly
absorbed in their environment. The following relationship was established empirically in order to implement
this effect:
0.8304
 OI 
ki ,OD =  i  , (12)
 OI sf 


where OIi = overlap integral of the absorbance spectrum of 2 % (w/v) of the respective UV filter at an optical
thickness of d = 0.002 cm and the irradiance spectrum of the lamp used for the photostability assessment [29],
and OIsf = overlap integral of the absorbance spectrum of the irregular sunscreen film and the irradiance spec-
trum of the lamp used for photostability measurements. If ki,OD < 1, there is a protective effect due to the overall
optical density. Only then ki,OD is considered by multiplying it with the decay constant of the respective filter.
For the SPF calculation with consideration of photoinstabilities the approach of Wloka et al. [28] is used.
For that purpose UV dose dependent concentrations of the UV filters are calculated using eq. (9) and with
that, UV dose dependent SPF values. The inverse of the UV dose-dependent SPF is plotted against the irra-
diation dose given in minimal erythemal doses (MEDs). The area under the curve can be interpreted as the
erythemally effective irradiation dose. When this area becomes unity, 1 MED has been transmitted through
the sunscreen film, what exactly corresponds to the principle of the in vivo SPF measurement. Thus the SPF
can be read at this point from the respective UV-dose given on the abscissa. An example with a resulting SPF
of 10 is shown in Fig. 8.
With photounstable UV filters, the spectrum of the absorbance of the sunscreen film varies with time.
According to Stanfield et al. [27] it is possible to determine the effective UV absorbance spectrum, regarding
the period from beginning of irradiation up to the point where 1 MED is transmitted. To this end, the transmit-
ted UV-doses with (Dp) and without protection (Du) are integrated over time, and transmittance, now called
integrated transmittance, is calculated for each wavelength λ:

Tint ( λ ) = Dp ( λ ) / Du ( λ ), (13)

0.15

0.10
1/SPF(t)

0.05

0.00
0 5 10 15
UV dose / MED

Fig. 8: Evaluation of the calculated SPF with photounstable sunscreens.

 B. Herzog and U. Osterwalder: Simulation of sunscreen performance 945

where
t MED
Du = ∫ Ss ( λ ) dt (14)
0


and
t MED
Dp = ∫ Ss ( λ ) ⋅ T ( λ,t ) dt . (15)
0


The integrated absorbance can then be obtained from integrated transmittance:

Eint ( λ ) = lg( 1 / Tint ( λ )). (16)

Figure 9 illustrates the concept of integrated transmittance and absorbance with an example of a pho-
tounstable composition (5 % EHMC, 4 % BMDBM, 3 % OCR) with calculated SPF of 17.5.

Calibration of the model

As already mentioned, the screw by which the calculated SPF can be adjusted to SPF in vivo data is the shape
parameter in eq. (4). At first sight, the model has only this single parameter. However, it is not possible to get sat-
isfactory correlation with compositions covering a brought range of sun protection factors using just one value
of the shape parameter. So the shape parameter has to vary as function of the filter composition, and there are
two observations indicating, how this could be explained: First, presence of a higher amount of emollient leads
to an increase of the SPF at constant filter composition, and second, there is a synergistic effect on the SPF when
sunscreens contain water- and oil-soluble UV filters rather than having the UV filters only in one phase of the
preparation. As the only input for the calculations is the UV filter composition, the first point can be handled by
scaling the shape parameter with the overall filter concentration (OFC) which for most sunscreen compositions
is an indicator of the amount of oil used. The overall UV filter concentration for filters i to n is given as:
n
OFC = ∑ i βi (17)

For the second point, a relevant parameter is the relative erythema active extinction in the oil phase,
REAE. For a single filter i with concentration βi the erythema active extinction EAEi is calculated via eq. (18):
400
EAEi = βi . ∑ λ = 290 E 11 ( i , λ ) ⋅ ser ( λ ) ⋅ Ss ( λ ) (18)


where E11(i, λ) is the specific extinction of filter i at wavelength λ, the other quantities have the same meaning
as in eqs. (1) and (2). The relative erythema active extinction in the oil phase is given via eq. (19):

1.5 1

1.2 0.8

No irradiation
Absorbance

Transmittance

0.9 0.6 Integrated spectrum

1 MED transmitted
0.6 0.4
No irradiation
Integrated spectrum
0.3 0.2
1 MED transmitted

0 0
290 400 290 400
Wavelength (nm) Wavelength (nm)

Fig. 9: Integrated spectra calculated for a sunscreen composition of 5 % EHMC, 4 % BMDBM, and 3 % OCR.
946 B. Herzog and U. Osterwalder: Simulation of sunscreen performance

REAE = ∑ EAEi (o) / [ ∑ EAEi (o) + ∑ EAEk (w) ], (19)

i i k

where EAEk(w) is the erythema active extinction of filter k in the water-phase. For the variation of the shape
parameter as function of filter composition, the following function has been found:

c = p 1 + p 2 ⋅ [ OFC p 3 /( OFC p 3 + p 4 )] + p 5 / 4 + p 5 ⋅ ( REAE − 0.5 ) 2 (20)

The values of the parameters have been adjusted to p1 = 0.33, p2 = 0.69, p3 = 0.89, p4 = 1.95, and p5 = 0.5.
Thus, the distribution of the UV filters in the water and the oil phase is taken into account as well as the
overall filter content, which in a way can be interpreted as formulation influences.
The calibration of the parameters in eq. (20) was achieved by using in vivo data of ring tests where several
laboratories had been involved, or our own data, where in vivo studies had been repeated several times on
the same sunscreen formulations (P1, P2 and P3 [6], S2 [30, 31], S1, C1, C2 and P4 [17, 32]). Table 3 lists the UV
filter compositions and corresponding data, and the correlation of simulated and in vivo data is shown in
Fig. 10. The slope of the correlation is 1.015 and the correlation coefficient is r2 = 0.9423.

Protection against UVA radiation

The in vivo method used for UVA protection assessment is persistent pigment darkening (PPD). PPD is the
official method for assessment of the UVA-protection in Japan [32]. In analogy to the erythemal response

Table 3: Calibration of the model parameters using sunscreens with in vivo data based on large sample sizes [6, 17, 30–32]; in
vivo data given with confidence intervals based on 95 % level of significance.

Sunscreen Filter composition SPF in vivo Calculated SPF

P1 2.7 % EHMC 4.2 ± 0.2 5.1

P2 7 % OD-PABA, 3 % B-3 12.7 ± 1.2 16.7
P3 2.78 % PBSA, 3 % EHMC, 0.5 % BMDBM 15.5 ± 1.5 12.1
S1 3 % EHMC, 5 % BMDBM 8.2 ± 0.8 9.8
S2 3 % EHMC, 5 % BMDBM, 3 % OCR, 2 % BEMT 16.0 ± 2.7 20.2
C1 3 % BEMT 7.7 ± 0.8 7.3
C2 2 % BEMT, 4 % MBBT, 5 % EHMC 29.0 ± 5.0 27.2
P4 10 % MBBT, 5 % EHMC 35.7 ± 3.2 36.2

40

Slope = 1.015

30 r2 = 0.9423
Calculated SPF

20

10

0
0 10 20 30 40
SPF in vivo

Fig. 10: Correlation of calculated SPF and in vivo SPF of formulations used for calibration of the model parameters (data in Table 3).
 B. Herzog and U. Osterwalder: Simulation of sunscreen performance 947

expressed by the SPF, the UVA-protection factor (UVAPF) values are determined by the biological endpoint of
spontaneous pigmentation after irradiation with a UVA light source.
The UVAPF is simulated in a similar way like the SPF from calculated transmittance T(λ) taking photo-
degradation of the filter molecules into account, but now employing the spectrum of an UVA lamp, SUVA(λ),
and the PPD action spectrum sPPD(λ) in eq. (8), and the time tMPD after which a minimal pigmenting dose
(1 MPD) is transmitted:
t MPD 400

UVAPF =
∑ ∑ 0 320
sPPD ( λ ) ⋅ SUVA ( λ )
(21)
t MPD 400
∑ ∑ s 0 320 PPD
( λ ) ⋅ SUVA ( λ ) ⋅ T ( t , λ )

For the calculation of UV transmittance in eq. (21), the same principles are applied as previously described
for the calculation of the SPF. However, the overall UV dose needed to achieve 1 MPD is higher than that to
achieve 1 MED. Thus, irradiation time with PPD experiments is about five times longer. Considering that the
UVB is filtered off, still a factor 3.2 has to be taken into account for the calculation of UV filter photodegradation.
An in vitro method for determination of the UVA-PF has been developed [13] and is meanwhile standard-
ized [14]. In order to simulate this method, the film irregularity model is not used, but a relationship between
parent film extinction and extinction of the film on the substrate, based on in vitro data, leading to similar
curves as shown in Fig. 7:

Ein vitro = a ⋅ E parent

b
(22)

Assuming the thickness of the homogenous parent film to be 0.00075 cm, the parameters were deter-
mined to a = 0.5072 and b = 0.5466. In that way in vitro extinction and transmittance spectra are obtained
resulting in a simulated in vitro SPF, the value of which may differ from the simulated in vivo SPF. The cor-
responding simulated in vitro transmission spectrum is then transformed, such that the simulated in vitro
SPF matches the simulated in vivo SPF. This is achieved by adjusting a parameter C as an exponent to the
transmittance T(λ):
400

UVAPFin vitro =
∑ 320
sPPD ( λ ) ⋅ SUVA ( λ )
(23)
400
∑ 320
sPPD ( λ ) ⋅ SUVA ( λ ) ⋅ T ( λ ) C

The irradiation step involved in this method [14], which scales with the UVAPF obtained without irradia-
tion, is also taken into account.
Other quantities characterizing the UVA protection such as UVA/UVB-ratio, critical wavelength and others
[15] can also be deduced from the calculated absorbance or transmittance spectra. The UVA/UVB-ratio is the
average absorbance in the UVA-range divided by the average absorbance in the UVB-range. The critical wave-
length (λc) is given as the wavelength at which 90 % of the total area under the absorbance curve is achieved.

Simulation examples

Variation of application amount

The question how the sun protection factor depends on application amount is of interest, since people tend
to apply much less sunscreen as is used in the in vivo SPF test performed on volunteers. A multi-center study
showed, that there is a linear dependence of the SPF on application amount [33], although some authors
claim an exponential relationship [34]. From eq. (1) it can be deduced that an exponential relationship can
only be expected if the transmittance would be approximately the same at all wavelength between 290 and
400 nm. As there is also significant erythemal efficacy in the UVA range, the SPF of a pure UVB sunscreen will
even show saturation-like behavior when its application amount is increased.
 948 B. Herzog and U. Osterwalder: Simulation of sunscreen performance

Figures 11 and 12 show calculations for two extreme examples, comparing a broad-spectrum sunscreen
(5 % BEMT and 5 % MBBT, UVA/UVB-ratio = 0.92) and a sunscreen containing only UVB filters (5 % EHT
and 5 % PBSA) UVA/UVB-ratio = 0.16), both with overall filter concentrations of 10 %. Indeed, the SPF of
the broad-spectrum composition shows slightly exponential behavior, whereas that of the UVB sunscreen
approaches saturation when the application amount is increased.
In [33] three sunscreen products were investigated in a ring study. All products showed linear dependence
of the SPF on application amount. In order to simulate the behavior of these products, the filter contents of
products B and C were analyzed. Product B with label SPF 20 contained 10 % OCR, 2.7 % BMDBM, 1.8 % TiO2,
1.0 % DTS, 1.0 % EHT, and 0.5 % TDSA (abbreviations are explained in the experimental section). Product C
with label SPF 25 contained 4.2 % EHT, 3.5 % TiO2, and 2.0 % BEMT. Knowing the UV filter concentrations it
is possible to calculate the SPF as function of application amount. The simulations shown in Fig. 13 are in
good agreement with the experimental data [33] and also represent approximately the linear dependence of
the SPF on application amount. The UVA/UVB-ratios are in between those of the extreme examples shown in
Fig. 12 (0.72 for product B, and 0.53 for product C).

2.0
UVA/UVB-ratio = 0.16
UVA/UVB-ratio = 0.92
1.5
Simulated film absorbance

1.0

0.5

0.0
290 345 400
Wavelength (nm)

Fig. 11: Calculated sunscreen spectra of irregular films of composition 5 % EHT + 5 % PBSA (UVA/UVB-ratio = 0.16) and
5 % BEMT + 5 % MBBT (UVA/UVB-ratio = 0.92).

40

UVA/UVB-ratio = 0.16

30 UVA/UVB-ratio = 0.92
Calculated SPF

20

10

0
0 0.5 1 1.5 2 2.5 3
Application amount (mg/cm2)

Fig. 12: Calculated SPF as function of application amount for two sunscreens with different UVA/UVB-ratios (0.16 and 0.92).
 B. Herzog and U. Osterwalder: Simulation of sunscreen performance 949

40
B, in vivo B, simulation
C, in vivo C, simulation
30
SPF

20

10

0
0 0.5 1 1.5 2 2.5
Application amount (mg/cm2)

Fig. 13: SPF as function of application amount for two sunscreens investigated in reference [33].

Variation of light source spectrum

For determination of SPF in vivo, a light source is used with a spectrum deviating in the UVA-I range
strongly from solar spectra, as the visible light is filtered off. The question how much this affects the
protection under natural solar conditions, has already been addressed in an experimental study [35].
It was observed that the SPF is lowered by about one third when UVA-I is not filtered. It was also dem-
onstrated, that with different types of solar simulators, SPF results may vary by more than a factor of
two [36]. Figure 14 shows a comparison of a standard solar spectrum [10], the spectrum defined in the
International SPF method [8] and the latter one multiplied with the transmission spectrum of the WG335
filter [37].
In Fig. 11 the irregular film absorbances of two quite different UV filter compositions are shown for the
usual application amount of 2 mg/cm2. Considering the SPF of the two compositions, the case with UVA/UVB-
ratio = 0.16 gives an SPF of 13.7 when calculated with the laboratory light source spectrum. Taking the spec-
trum of the standard sun, the SPF is only 10.6, about 23 % lower. In comparison, the SPF of the composition
with UVA/UVB-ratio = 0.92 is 26.3 with the laboratory lamp and 25.0 when taking the standard sun spectrum,
which is only 5 % lower. That means, the more uniform a sunscreen absorbs in the UVB and the UVA range,
the less its performance will depend on the spectrum of the light source.

1.2

1.0
Irradiance (W/m²/nm)

0.8

0.6

Standard sun
0.4
Source intl. meth.
0.2 Source intl. meth.
with WG335
0
290 310 330 350 370 390
Wavelength (nm)

Fig. 14: Different light source spectra.

950 B. Herzog and U. Osterwalder: Simulation of sunscreen performance

Table 4: Correlation of in vivo and calculated SPF of formulations using a UV filter concentrate [39, 40]; in vivo data given with
confidence intervals based on 95 % level of significance.

% Premix Filter composition SPF in vivo Calculated SPF

6 1.2 % EHMC, 1.2 % DHHB, 0.36 % EHT, 0.3 % BEMT 6.5 ± 2.5 6.1
10 2 % EHMC, 2 % DHHB, 0.6 % EHT, 0.5 % BEMT 10.0 ± 2.6 9.8
15 3 % EHMC, 3 % DHHB, 0.9 % EHT, 0.75 % BEMT 17.6 ± 5.0 14.8
20 4 % EHMC, 4 % DHHB, 1.2 % EHT, 1 % BEMT 20.9 ± 4.8 19.8
25 5 % EHMC, 5 % DHHB, 1.5 % EHT, 1.25 % BEMT 30.4 ± 3.8 25.0
30 6 % EHMC, 6 % DHHB, 1.8 % EHT, 1.5 % BEMT 35.0 ± 5.3 30.2
50 10 % EHMC, 10 % DHHB, 3 % EHT, 2.5 % BEMT 48.8 ± 4.9 51.0

60

50
Slope = 0.944

r2= 0.9680
40
Calculated SPF

30

20

10

0
0 10 20 30 40 50 60
SPF in vivo

Fig. 15: Correlation of calculated SPF and in vivo SPF of formulations based on a UV filter concentrate (data in Table 4).

Development of new sunscreens

The sunscreen simulator model described here [38] is widely used in the cosmetic industry for the design of
new sunscreen formulations. An example where it has been successfully applied was the development of
a premix concept for a simple preparation procedure of sunscreens [39, 40]. The concentrate consists of a
mixture of four UV absorbers, 20 % EHMC, 20 % DHHB, 6 % EHT, and 5 % BEMT, and cosmetic oils and emul-
sifiers added to 100 %. Upon dilution with water an emulsion forms spontaneously. In that way sunscreen
formulations can be easily prepared. With the chosen filter composition the resulting emulsions comply with
the requirements for UVA protection, such as UVAPF/SPF ≥ 1/3 and λc ≥ 370 nm. In Table 4 filter concentra-
tions, in vivo SPF and calculated SPF data are given for different dilutions of the concentrate. The correlation
is shown in Fig. 15. The model prediction of the SPF (slope = 0.944) and the correlation of in vivo and simu-
lated SPF data (r2 = 0.9680) are excellent.

Conclusion
For the simulation of sunscreen performance the following elements are needed: a database with quantita-
tive UV spectra of the relevant UV filters, a mathematical description of the irregularity profile of the sun-
screen film on the skin, and the consideration of changes in UV filter concentration due to photoinstabilities.
 B. Herzog and U. Osterwalder: Simulation of sunscreen performance 951

Such simulations can significantly reduce the amount of experimental work needed when developing new
sunscreen formulations, and additionally help understanding how sunscreens work when simulating their
behavior under different conditions [41].

References
[1] F. R. de Gruijl. Skin Pharmacol. Appl. Skin Physiol. 15, 316 (2002).
[2] H. N. Ananthaswamy, S. M. Loughlin, S. E. Ulrich, M. L. Kripke. J. Invest. Dermatol. Symp. Proc. 3, 52 (1998).
[3] F. R. de Gruijl, H. J. C. M. Sterenborg, P. D. Forbes, R. E. Davies, C. Cole, G. Kelfkens, H. van Weelden, H. Slaper, J. C. van der
Leun. Cancer Res. 53, 53 (1993).
[4] S. Seité, D. Moyal, M. P. Verdie, C. Hourseau, A. Fourtanier. Photodermatol. Photoimmunol. Photomed. 16, 3 (2000).
[5] R. Schulze. Parfüm. Kosmet. 37, 6, 7, 310, 365 (1956).
[6] J. Ferguson, M. Brown, D. Alert, S. Bielfeldt, J. Brown, A. Chardon, C. Hourseau, C. Mazilier, J. Cuthbert, D. Arcy-Burt,
J. ­Jolley, M. Murdoch, P. Finkel, P. Masson, F. Merot, A. Maclennon, J. Poret, S. Siladji. Int. J. Cosmet. Sci. 18, 203 (1996).
[7] Food and Drug Administration, FDA, USA. Labeling and effectiveness testing; sunscreen drug products for over-the-counter
human use, final rule. Fed. Reg. 76/117, 35620 (2011).
[8] International sun protection factor (SPF) test method CFFA SA, COLIPA, JCIA, CTFA, (2006).
[9] ISO 24444:2010. In vivo determination of the sun protection factor (SPF).
[10] B. L. Diffey, J. Robson. J. Soc. Cosmet. Chem. 40, 127 (1989).
[11] H. Tronnier, D. Kockott, B. Meick, N. Hani, U. Heinrich. Parfüm. Kosmet. 77, 326 (1996).
[12] S. Miksa, D. Lutz, C. Guy. Cosmet. Toiletries 129, 30 (2014).
[13] COLIPA method for the in vitro protection of UVA protection provided by sunscreen products. The European Cosmetics
Toiletry and Perfumery Association – COLIPA (now: Cosmetics Europe), Rue de la Loi 223/2, B-1040 Bruxelles (2007).
[14] ISO 24443:2012. Determination of sunscreen UVA photoprotection in vitro.
[15] B. L. Diffey. Int. J. Cosmet. Sci. 16, 47 (1994).
[16] B. Herzog. J. Cosmet. Sci. 53, 11 (2002).
[17] B. Herzog, S. Mongiat, K. Quass, C. Deshayes. J. Pharm. Sci. 93, 1780 (2004).
[18] B. Herzog, M. Wehrle, K. Quass. Photochemistry and Photobiology, Symposium in Print: Pharmaceutical Photochemistry
85, 869–878 (2009).
[19] R. M. Sayre, P. P. Agin, G. J. LeVee, E. Marlowe. Photochem. Photobiol. 29, 559 (1979).
[20] A. F. McKinlay, B. L. Diffey. CIE J. 6, 17 (1987).
[21] J. J. O’Neill. J. Pharm. Sci. 73, 888 (1984).
[22] B. Herzog, in Sunscreens – Regulation and Commercial Development, Nadim Shaath (Ed.), 3rd ed., p. 881, Taylor & Francis,
Boca Raton (2005).
[23] L. Ferrero, M. Pissavini, S. Marguerie, L. Zastrow. J. Cosmet. Sci. 54, 63 (2003).
[24] B. Herzog, in Colloids in Cosmetics and Personal Care, TF Tadros (Ed.), vol. 4, p. 275, Wiley VCH, Weinheim (2008).
[25] R. T. Thomas, in Rough Surfaces, 2nd ed., p. 91, Imperial College Press (1998).
[26] E. P. Abbott, F. A. Fireston. Mech. Eng. 55, 569 (1933).
[27] J. Stanfield, U. Osterwalder, B. Herzog. Photochem. Photobiol. Sci. 9, 489 (2010).
[28] M. Wloka, R. F. M. Lange, H. Flösser-Müller. Proc. Int. Sun Protection Conference, London (2005).
[29] B. Herzog, S. Müller, M. Sohn, U. Osterwalder. SOFW J. 133, 26 (2007).
[30] D. Moyal, M. Pissavini, F. Boyer, V. Perier, J. H. Frelon. Int. J. Cosmet. Sci. 29, 443 (2007).
[31] The in vivo SPF of the S2 standard sunscreen was provided by D. Moyal, L’Oreal Recherche.
[32] JCIA Measurement Standard for UVA Protection Efficacy. Japan Cosmetic Industry Association – JCIA, 9-14, Toranomon
2-Chome, Minato-Ku Tokyo, p. 105 (1995).
[33] R. Bimczok, H. Gers-Barlag, C. Mundt, E. Klette, S. Bielfeldt, T. Rudolph, F. Pflücker, U. Heinrich, H. Tronnier, W. Johncock,
B. Klebon, H. Westenfelder, H. Flößer-Müller, K. Jenni, D. Kockott, J. Lademann, B. Herzog, M. Rohr. Skin Pharmacol.
Physiol. 20, 57 (2007).
[34] N. Bech-Thomsen, H. C. Wulf. Photodermatol. Photoimmunol. Photomed. 9, 242 (1993).
[35] R. M. Sayre, J. Stanfield, A. J. Bush, D. L. Lott. Photodermatol. Photoimmunol. Photomed. 17, 278 (2001).
[36] B. Uhlmann, T. Mann, H. Gers-Barlag, D. Alert, G. Sauermann. Int. J. Cosmet. Sci. 18, 13 (1996).
[37] http:\\www.uqgoptics.com/materials_filters_schott_colourless_WG335.spx. Accessed 31 August 2015.
[38] BASF sunscreen simulator, BASF SE, Ludwigshafen, Germany. 2010. Available at http://www.basf.com/sunscreen-simulator.
[39] S. Acker, M. Hloucha, U. Osterwalder. SOFW J. 40, 24 (2014).
[40] S. Acker, M. Hloucha, U. Osterwalder. Global Ingredients and Formulation Guide, SOFW, p. 152 (2014).
[41] D. J. McGarvey. Educ. Chem. 44, 116 (2007).