AIChE Journal Formulating Sunscreen 2011
AIChE Journal Formulating Sunscreen 2011
AIChE Journal Formulating Sunscreen 2011
Component
Elisa Conte and Raqul Gani
Dept. of Chemical and Biochemical Engineering, CAPEC, Technical University of Denmark, Lyngby DK-2800, Denmark
Yuen S. Cheng and Ka M. Ng
Dept. of Chemical and Biomolecular Engineering, The Hong Kong University of Science and Technology,
Clear Water Bay, Kowloon, Hong Kong
DOI 10.1002/aic.12582
Published online March 8, 2011 in Wiley Online Library (wileyonlinelibrary.com).
A systematic methodology for the design and verication of chemical-based products
is proposed. By integrating modeling, and experiments, the search space is efciently
scanned to identify the feasible product candidates. The product design (or verica-
tion) problem consists of three stages: computer-aided design (Stage 1), which gener-
ates a list of feasible candidates, experimental planning (Stage 2), which generates a
list of experiments and checks the available experimental set-ups, and experimental
testing (Stage 3), which measures the necessary data and veries the desirable attrib-
utes of the nal product. The rst stage (Stage 1) has been covered in previous publi-
cations, along with detailed case studies. The development of Stage 2 and Stage 3 is
considered in this article and highlighted through two case studies involving the design
and validation of an insect repellent lotion and a sunscreen lotion. V VC
2011 American
Institute of Chemical Engineers AIChE J, 58: 173189, 2012
Keywords: product design, formulation design, experimental testing, insect repellent
lotion, sunscreen lotion
Introduction
In the last years, the chemical industry has been widening
its horizons toward higher value-added materials.
1,2
This
change is not simply a change from commodities to special-
ties, but it signies a paradigm shift from focus on material
purity to material performance. In chemical product design,
the problem is to nd the most appropriate combination of
chemicals and the processing technology that will lead to the
desired product attributes set by market demands.
The development of systematic methodologies, tools, and
strategies for product design is crucial in a highly competi-
tive commercial environment.
3,4
It is generally agreed that
product design is a mix of skills from different disciplines
such as business, social sciences, and ne arts, not only
chemical sciences and chemical engineering.
5,6
To meet this need, various attempts have been made to
develop systematic methodologies for the design and devel-
opment of chemical-based products. Cussler and Moggridge
7
suggested a four-step procedure based on needs, ideas, selec-
tion, and manufacture. According to the perspective of
Gani,
8,9
chemical product design is the discipline that guides
the developer in identifying the most appropriate chemical(s)
that will exhibit and/or impart the desired behavior. Wester-
berg and Subrahmanian
5
recommended organizing the design
procedure. Wibowo and Ng
10
developed a procedure for the
manufacture of creams and pastes, which was subsequently
generalized for chemical-based consumer products.
11
Fung
and Ng
12
presented a systematic procedure for the manufac-
ture of tablets and capsules. Cheng et al.
13
formulated an
approach, which involves business and management deci-
sions in product development.
Correspondence concerning this article should be addressed to R. Gani at
rag@kt.dtu.dk.
V VC
2011 American Institute of Chemical Engineers
AIChE Journal 173 January 2012 Vol. 58, No. 1
In addition to the general approaches described above, a
large number of computer-aided methods have also been
developed. Computer-aided methods have been developed for
molecular design,
14
solvent design,
1518
mixture design,
19
polymer design,
2023
and refrigerant design.
24,25
The algo-
rithms employed can be classied under the following types:
generate and test algorithm,
15,17,24,26
genetic algorithm,
22
mathematical programing,
16,21
component-less design techni-
ques,
27,28
combinatorial optimization,
29
and hybrid methods.
14
The design and verication of formulated products
In the chemical industry, several chemicals are often
combined in formulations to obtain products with different
functions and qualities. Many household products such as
cosmetics, pharmaceuticals, and personal care products are
formulations. This article focuses on lotions which are for-
mulations in liquid form.
Lotions are usually constituted of one or more active
ingredients (AIs) for providing the main product activities,
a mixture of solvents serving as the delivery system of the
AIs to the desired surface, and additives (usually present in
small concentrations, less than 2% in volume) for enhanc-
ing the end use product properties. The solvent mixture
evaporates after application of the product. Solvent selec-
tion and solvent mixture design are key activities in the
design of such a product, since many of the end-use prod-
uct qualities such as drying time, toxicity, product form
and so on are closely related to the solvent/solvent mixture
present in the product.
A systematic methodology for the design and the veri-
cation of formulations with a liquid form has been recently
developed,
30,31
design and verication of formulated prod-
ucts are two different scenarios the designer may encoun-
ter. When designing a completely new product,
30
the iden-
tities of the ingredients of the formula are not known; the
objective is to identify the formulation of chemicals that
matches the a priori dened criteria. In the verication of
formulated products,
31
the identities of most of the chemi-
cals are known, and/or a shortlist of candidate ingredients
is given (avoiding thereby the generation/screening of
numerous alternatives). The aim here is to verify the
(liquid) phase stability of the actual formula and/or to cal-
culate its chemical and physical properties to check the
product performance.
The methodology proposed in this work for the design
and the verication of formulated products with a liquid
delivery system strikes a balance between model-based and
experimental approach to product design. Table 1 compares
the model-based and experimental approaches in terms of
the tools used, objectives, research environment, and uncer-
tainty. The model-based approach aims at screening a large
number of product alternatives to identify a small number of
candidates (design scenario), or at verifying the mutual solu-
bility and the functions of the chemicals present in a given
shortlist (verication scenario) for further testing, validation,
and modication through experimental research. The accu-
racy of predictions depends on the reliability of mathemati-
cal models and the assumptions and hypothesis taken. In
addition, modeling cannot cover sensorial properties such as
the turbidity, color, scent, greasiness, and stickiness of the
product although such properties can be easily determined
by experiments. Any experiment-based trial-and-error
approach is very demanding in terms of time and resources.
It also requires experience in the form of heuristics, guide-
lines, etc., to adjust the product to meet the target values of
its attributes.
By combining modeling with experiments, the uncertain-
ties of the model-based approach are compensated by the
experimental tests whereas the number of experiments is
reduced through model predictions. The proposed methodo-
logy based on the integrated approach consists of three
stages, as shown in Figure 1. In Stage 1, computer-aided
design reduces the search space and provides a list of poten-
tial candidates. Stage 2 involves planning for experimental
identication of the important product attributes, and to put
together the experimental set-up. In Stage 3, experiments are
performed. Modications and improvements of the base case
formula are made, if necessary, to meet the target values of
the product attributes.
This article deals with the design of formulated products,
and attention is focused on Stage 2 and Stage 3 of the for-
mulation design methodology. Stage 1, involving computer-
aided liquid formulation design has already been presented
in Conte et al.
30
whereas details of computer-aided verica-
tion of formulated products can be found in Conte et al.
31
Two case studies are used in this article to illustrate the
workow: the design of formulations for an insect repellent
and a water-resistant sunscreen lotion.
An Integrated Methodology
for Formulation Design
Before describing Stage 2 and Stage 3 (the focus of this
article) of the methodology in detail, Stage 1 related to the
design scenario is briey reviewed below.
Table 1. Comparison Between Model-Based and Experimental-Based Approach for Product Design
Model-Based Experimental-Based
Tools Mathematical predictive models Knowledge base, experience of few expertises
Objective Screen numerous candidatesobtain a base case Manufacture the end-use product
Environment Virtual reality Reality
Uncertainty Models uncertainty, assumptions, hypothesis No uncertainty
Advantages Time and resources are spared Manufacture of end-use products
Disadvantages High uncertainty, necessity of an experimental
validation, some properties can not be modeled
(scent, appearance,), models are limited to
some kind of chemical
Long development times, high consumption
of resources, necessity of the knowledge base
174 DOI 10.1002/aic Published on behalf of the AIChE January 2012 Vol. 58, No. 1 AIChE Journal
Stage 1: Computer-aided design
Stage 1 is based on the reverse design technique
9
: the
methodology guides the user through the design of a formu-
lation that matches the a priori dened targets. The targets
are set during the problem denition (task S1-D1). AIs are
identied (task S1-D2) through database and knowledge base
search (patents, literature, company knowledge, and so on).
Information about toxicity, efciency, and concentration that
ensures the desired efcacy is also gathered. If measured
values of the properties of the AI chemicals are not avail-
able, they are predicted through group contribution meth-
ods.
32
The identication of the pure solvent or the design of
a solvent mixture (task S1-D3) is an important step as it has
to guarantee the complete dissolution of the AIs and the
phase stability of the formula, in addition to conferring the
desired properties to the product. The models used for
screening are usually simple linear models, but rigorous
models which account for the mixture excess properties can
also be employed, if necessary. The design algorithm pro-
duces a list of feasible candidate solvent mixtures, but one
optimal candidate is identied in the optimization sub-task.
In task S1-D4, additives are chosen for the identied base
case formula. No iteration is required between the different
tasks of Stage 1.
The computer-aided stage produces a base-case formula
that serves as the starting point for verication and nal
selection by experiments. The base-case formula is just an
indication of which ingredients should be present in the for-
mulation, and in which ratios, but experiments are extremely
important to amend the formula and produce the product
prototype.
The computer-aided stage of the methodology has been
implemented within a computer-aided framework.
30,31,33
A
number of methods and tools, as well as property models,
databases, and knowledge base have been included in the
framework.
30
The methods (and the associated computer pro-
grams) are:
1. The mixture classication algorithm, for the classica-
tion of mixture according to hydrogen-bonding properties
and, therefore, the contribution of excess properties to the
mixing process.
2. The MIXD algorithm, for the design of single liquid
phase binary mixtures of solvents. MIXD contains also the
STABILITY algorithm.
3. The STABILITY algorithm, for the stability test of bi-
nary solvent mixtures.
The developed databases contain data and information for
several kinds of AIs, solvents, and additives.
The knowledge base contains data, information, and rules
retrieved from literature, patents, patented products,
insights, and common sense, which guides the designer
through the design (or verication) of formulated products.
For instance, the knowledge base contains information on
how to identify the performance criteria/consumer needs/
product attributes usually required for some specic formu-
lated products, and on how to translate these criteria into
technical specications (target properties and numerical
constraints).
As stated in Part I,
30
this work does not aim to determine
the compositions of AIs and additives in the formulation,
since the necessary models are not available for this task.
Instead, values found in the literature (patents) or in real
product recipes have been used (knowledge base). Note,
however, the amount and compositions of the solvent mix-
ture are calculated.
Figure 1. Work-ow diagram for the integrated meth-
odology for formulation design.
AIChE Journal January 2012 Vol. 58, No. 1 Published on behalf of the AIChE DOI 10.1002/aic 175
Stage 2: Experimental planning
Stage 2 serves as the link between the computer-aided
screening of alternatives and the experimental validation.
Here, the availability of the product ingredients and of the
needed experimental set-ups is conrmed. If one of the
chemicals is not available, the loop L1 in Figure 1 is fol-
lowed to nd suitable alternative chemicals. For instance, if
one of the solvents of the base case formula is not available,
the second best solvent mixture identied by the mixture
design algorithm could be taken into consideration during
experimental planning. If an experimental set-up is not avail-
able, alternative solutions need to be found to measure the
corresponding property. At the end of Stage 2, a list of
experiments that are ready to be performed is produced.
For lotions, the experiments are divided into three levels.
First, the AI/AIs should be tested, if necessary, to verify if
they possess the needed functional properties for the prod-
uct. For instance, if an AI is chosen as an insect repellent
for tropical areas, it should be proven that it can repel the
mosquitoes that are present in tropical areas. The second
level involves testing of the solvents and the solvent mix-
ture. The solvents should be able to dissolve the AI/AIs
without phase separation. The physical/chemical properties
of the solvents/solvent mixture which are considered critical
for the product under consideration should be measured as
well. Since the additives are present in low concentration,
their effect on the nal product properties is expected to be
negligible. Therefore, they are not tested individually but
the effect on the overall formulation is observed in the
next level. Availability and price of these materials should
also be considered. The third level is the experimentation
on the prototypes. For lotions, the production of the proto-
types is simply mixing. The overall formula has to meet
the a priori dened targets. Spray-ability for the spray
products and spread-ability for products such as paints have
to be tested. Some tests will include also the validation/
measurements of properties that cannot be modeled such as
the sensorial factors (color, turbidity, scent, greasiness,
etc.), shelf life and pH, and properties such as phase sepa-
ration at extreme temperatures that have not been consid-
ered in the computer-aided design.
All the properties considered during the computer-aided
design stage (Stage 1) have to be experimentally veried in
Stage 3. In some cases, it is not necessary to measure a
particular property, if it is easier to perform a related
experiment. For example, instead of measuring the solubility
parameters, it is easier to measure solubility and also con-
rm miscibility. In other cases, it is not possible to perform
some experiments due to the lack of experimental set-up
and/or chemicals. In this case, substitutive experiments can
be performed.
Stage 3: Experimental validation
Stage 3 consists of two iterative loops. In the inner loop,
the simple and non time consuming tests are carried out
(task S3-1). If not all the tests give satisfactory results, prob-
lems are identied (task S3-2) and modications are made to
x the problems (task S3-3). This loop is iterated until all
the simple/fast tests give satisfactory results; that is, when
all the a priori dened constraints are satised. When this
happens, task S3-4, in the outer loop, is performed. Here, all
the long time duration and difcult tests are carried out. If
any of these tests do not give satisfactory results, the prob-
lems are identied and modications in the formulation are
suggested (tasks S3-2 and S3-3). The simple tests may have
to be performed again since the modications could affect
the product properties considered in the inner loop (simple
tests and measurements). This is continued until a formula
that satises all the targets has been identied, and process
synthesis and scale-up can begin (these two last steps are not
considered in this work).
The computer-aided stage and the experimental stage are
not independent of each other. Some of the problems found
in the experimental stage can be solved by replacing one or
more ingredients. Thus, it becomes necessary to go back to
the computer-aided stage during task S3-3 (amendments),
following the iterative loop L2 (Figure 1), which links Stage
3 back to Stage 1 of the methodology.
Case Studies
Two case studies are presented here to highlight the ex-
perimental validation related stages of the integrated method-
ology workow: a water-based insect repellent and a water-
proof sunscreen, both products in the form of a spray. The
development of the base case formula through the computer-
aided design stage of the methodology (Stage 1) is not
described in detail, since the reader can refer to previous
publications
30,34
for more explanations. Instead, only a sum-
mary of the results from this stage is reported here.
Design of a water-based insect repellent
The aim of this case study is to design an insect repellent
lotion. Water should be one of the formulation ingredients,
because of safety and cost concerns. Some well known
insect repellents in the spray form are based on wateralco-
hol mixtures, like the well known product from Bayer,
Autan
VR
. The market for consumption is non tropical areas
(such as Europe).
Computer-Aided Stage (S1-D). Task S1-D1: Problem
denition. Consumers want a product with the following
characteristics: high effectiveness against mosquitoes (the
main function of the product); high compatibility with other
materials (fabrics, plastics, etc); a water-based product, for
safety and toxicology reasons; good sensorial factors and cos-
metic properties, that is, nice odor, appearance, and good skin
feeling; low price; long durability (it should not be needed to
apply the product often during exposure to mosquitoes); low
toxicity; high stability (no separation of phases); good user
friendliness, such as a spray product; long shelf life. The
effectiveness of the product is guaranteed by choosing suita-
ble AIs. Material compatibility is guaranteed by selecting non
aggressive and corrosive chemicals. The scent of the product
can be improved by adding perfumes (additives). The durabil-
ity of the product is related to the evaporation time T
90
; the
toxicity is related to the parameter LC
50
; the stability of the
product can be controlled through the solubility parameter d,
the Gibbs energy of mixing and the tangent plane distance
30
(DG
mix
, TPD); the spray-ability is related to the kinematic
176 DOI 10.1002/aic Published on behalf of the AIChE January 2012 Vol. 58, No. 1 AIChE Journal
viscosity m and the density q (or molar volume V). Table 2
shows the lower and upper bounds of the constraint values for
the target properties described above.
Task S1-D2: AI identication. Only one AI is necessary,
since the main function of the product is only one: to repel
mosquitoes. DEET,
35,36
natural AIs,
37
and Picaridin
35
are
usually employed as AIs in insect repellents. Picaridin was
selected because of its superior qualities. Unfortunately, Pic-
aridin has a very low solubility in water (8.6 Kg/m
3
at 273
K) and a second solvent (an alcohol) has to be added to
form a solvent mixture which can dissolve Picaridin (Picari-
din shows high alcohol solubility). For this reason, it is
necessary to verify whether the solubility parameter of the
second solvent in the mixture (besides water) is close to that
of Picaridin (d
AI
3 MPa
1
=2
), substituting the constraint on
the solubility parameter of the solvent mixture (see Table 2)
with the constraint of Eq. 1.
d
AI
3:0 _d
2
_d
AI
3:0 21:1 _ d
2
_ 27:1 (1)
d
2
is the solubility parameter of the second solvent present in
the solvent mixture (besides water). The Hildebrand solubility
parameter of Picaridin (d
AI
) corresponds to 24.1 MPa
1
=2
,
calculated with the M&G GC
Stability d, DG
mix
, TPD Yes Exps. 1, 2, 3, 4, and 10
Usually, toxicity values are taken from MSDS in experiment-based product design.
Experiment 10 veries stability in a range of temperatures around 300 K, whereas in stage 1 only one temperature was used for the design (300 K).
Table 4. First (Base Case) and Second (Final) Iteration
Formulations for the Insect Repellent Case Study
Family Chemical
% w
i
Base Case
(First Iteration Formula)
Second Iteration
Formula
AI Picaridin 10.00 9.69
Solvent
mixture
2-Propanol 45.65 44.25
Water 43.35 42.01
Additives Linalool 1.00 4.00
Acetic acid 0.05
178 DOI 10.1002/aic Published on behalf of the AIChE January 2012 Vol. 58, No. 1 AIChE Journal
solution was left to rest for at least 18 h, to give enough
time for a possible phase separation. The experiment was
terminated when phase split occurred.
Other solubility verications: they were performed by
mixing and observing if phase stability occurred.
Density measurements: a known volume of liquid at
300 K (design temperature) was weighed.
Viscosity measurements: a Brookeld viscosimeter
(model DV-II Pro, adaptor UL/Y, spindle zero) was
employed. The temperature was controlled with a thermal
bath since the adaptor had a jacket.
Evaporation time measurements: the T
90
values (for
pure compounds) employed in the computer-aided design
were data (or predictions through a correlations based on
such data) measured with the shell thin lm evaporometer
according to the standard method ASTM D3539-87.
41
Such
an apparatus was not available in the laboratories where
the experiments were performed. An alternative apparatus
was used for the T
90
measurements: a qualitative lter
paper (from Advantec) of 7 cm diameter was leaned on a
Petri glass dish and introduced in a closed precision digital
scale (to exclude any noises affecting evaporation); 0.05 ml
of chemical was spread with a syringe on the lter paper,
creating a circle of about 2.6 cm of diameter; the weight
change was recorded along time. The percentage of weight
loss was plotted against the time (seconds) and a trend was
generated, from which the T
90
could be calculated. Since a
different apparatus for the T
90
measurements was employed,
results could not be compared with the values predicted
during the computer-aided design. But the evaporation
trends gave useful information about the way the formula-
tion evaporated and comparison between trends could be
performed to understand how the single chemicals affected
the formula evaporation.
Spray-ability verication: this test was performed by
spraying the lotion though a nozzle.
Cosmetic properties verication: greasiness, stickiness,
and irritating power of the formula was evaluated by apply-
ing the product on the skin.
pH measurement: indicator strips (Merck) were
employed.
Verication of the stability of the formula at tempera-
tures other than 300 K: this verication was performed by
storing one product sample in a fridge at a temperature of
278 K and another sample in an oven at a temperature of
318 K, for at least 1 week.
Shelf life evaluation: a product sample was left to rest
at room temperature for three months and any change in
appearance, odor, and consistency, as well as stability of the
formula was checked.
Experimental Validation Stage (S3). All results from this
stage are summarized in Table 6.
Task S3-1, iteration 1: Simple/fast tests. Experiment 1: The
solubility of Picaridin in water was measured to be 9.3 g/l
between 273 K and 275 K. This is in reasonable agreement
with the literature value of 8.6 g/l at 273 K.
Experiments 24: The mixture isopropanolwater was
found to be stable; Picaridin, as well as linalool, could be
dissolved in the solvent mixture. The product formula was
found to be a single liquid phase.
Experiment 5: Table 7 summarizes the values used in the
computer-aided design and the measured values for density,
molar volume, dynamic viscosity, and kinematic viscosity (the
kinematic viscosity was not measured, but calculated using the
experimental values of dynamic viscosity and density).
Properties of linalool were not measured because the
amount of chemical available was not sufcient and very ex-
pensive. The values used for water in the computer-aided
design refer to water that was not treated with lters whereas
the water used in the experiments was deionized water.
The estimated viscosity of Picaridin was found to be quite
different from the experimental value. In fact, the M&G model
employed for the estimation of the viscosity
42
had been shown
to work very well with small molecules such as solvents but it
had not been tested before for multifunctional molecules with
complex molecular structures, like Picaridin. Thus, the M&G
model parameters need to be ne-tuned for AIs.
The experimentally measured values are reliable since
they correspond to values published by others.
43
The fact
that the predicted values for the viscosity (see Table 3) is far
away from the measured ones is due to the difculty in pre-
dicting the viscosity for alcoholwater mixtures, as already
mentioned by Wu,
44
where it was also suggested to ne-tune
the UNIFAC-model parameters with viscosity data (this has
not been done in this work). It can be noted, however, that
the experimentally measured values of the mixture
Table 6. Experimental Results for the Insect Repellent Case Study: First (Base Case) and Second (Final) Iteration
No. Experiment ResultFirst Iteration ResultSecond Iteration
1 Solubility limit of Picaridin in water Low solubility (9.3 g/l @ 2023
C)
2 Phase stability of the solvent mixture Successful
3 Solubility of AI in the solvent mixture Successful
4 Solubility of linalool in the solution Picaridin
solvent mixture
Successful
5 m and q of pure solvents, solvent mixture, and formula Matching a priori dened constraints* Matching a priori dened
constraints
6 T
90
of pure compounds, solvent mixture, and formula Satisfactory Satisfactory
7 Formula sprayability Successful Successful
8 Appearance (turbidity/color), odor, stickiness,
greasiness, and irritation
Not satisfactory (too strong scent of
Picaridin), too sticky
Reduced acceptable
stickiness
9 pH Not satisfactory Satisfactory
10 Stability at different temperatures than 300 K (278
and 318 K)
Successful
11 Shelf life Successful
*Large deviation between predicted and experimental values for Picaridin viscosity.
AIChE Journal January 2012 Vol. 58, No. 1 Published on behalf of the AIChE DOI 10.1002/aic 179
viscosities still matched the viscosity constraint, and did not
make the product infeasible.
Experiment 6: The evaporation time measurements were
performed at room temperature of 295 K and at a humidity
of 47%. Figure 2a shows the trends of the percentage of
weight lost during the evaporation vs. the time for pure sol-
vents, solvent mixture, and overall formulation (base case).
The trends of Figure 2a reveal the reason why existing for-
mulations are not based just on water or do not show high
water concentration: the time for complete evaporation of
water is too long. The solvent mixture trend is hidden by the
formulation trend since they are almost overlapping, but
while the solvent mixture reaches 90% of weight loss in
around 51 min, the overall formula reaches 90% of weight
loss in almost 2.5 h.
The pure compounds show smooth trends and just before
reaching the 90% of loss the trends suddenly change their
slopes. Instead, the solvent mixture and the formulation
trends show quite premature changes in the slopes. Both
the solvent mixture and the formulation trends show a clear
inection at around 500 s (4050% of weight loss): this is
the point at which almost all the isopropanol had evapo-
rated. Just after almost all the isopropanol had evaporated
the water started to evaporate too. At around 2700 s and
8590% of weight loss, there is a second inection in the
formulation trend. This is the point at which almost all the
solvent mixture had evaporated. The rest of the trend
reached the 90% of loss almost asymptotically, employing
a very long time, since 11 wt % of the overall formula is
constituted of Picaridin and linalool (which have low vapor
pressures therefore long evaporation times). After this anal-
ysis, it can be concluded that the formulation evaporated as
desired: the solvent mixture evaporated at rst whereas the
AI and the additive stay on the desired surface for longer
time, providing the desired activity and a high product du-
rability.
Experiment 7: The formula could be sprayed through a
commercial nozzle. Spray-ability was veried.
Experiment 8: The formula was completely transparent.
The sensorial feeling on the skin after evaporation was a
slight stickiness, caused by Picaridin. No greasiness was felt,
nor irritation. The scent was not pleasant since the Picaridin
odor was still dominant.
Experiment 9: The pH of the formula was measured. The
pH was 8.5, too high for a personal care product (a pH
between 5 and 7.5 is preferred, since the skin pH is 5.5) and
it could cause irritation to some sensitive skin types.
Tasks S3-2 and S3-3, iteration 1: Problem identication and
amendments. The problems encountered were the unpleas-
ant scent of the formula and the high pH value.
The modications suggested were two. At rst, linalool
Figure 2. Trends of weight loss (percentage) vs. time
(seconds) for pure solvents and solvent mix-
ture. (a) First iteration formula (base case)
and (b) second iteration formula (nal for-
mula) for the insect repellent lotion.
Table 7. Property Values Used in the Computer-Aided Design (est) and Values Measured with Experiments (exp) for Pure
Compounds and Solvent Mixture for the Insect Repellent Case Study
Property
Picaridin Isopropanol DI Water Solvent Mixture Formulation
exp est exp est exp est exp
est
Base Case (First Iteration) Second Iteration lin rig
q 1066.8 1070.0* 807.4 782.5* 965.4 1000.0* 902.7 874.8 952.7 944.7
V 215.0 214.3* 74.4 76.8* 18.6 18.0* 31.13 32.13 32.62 33.71
g 76.3 44.60 2.13 2.06* 1.02 0.89* 2.99 1.15 1.16 3.80 4.27
m 71.5 41.68 2.64 2.63 1.06 0.89 3.31 1.31 1.33 3.99 4.52
Experimental values for the formulations (base case and second iteration) are also shown.
Units of measure: q (kg/m
3
), V (10
3
m
3
/kmol), g (mPa s
1
), m (10
6
m
2
/s).
Lin stands for linear model (linear mixing rule) and rig for rigorous model (viscosity calculation).
38
*The value used in the calculation was experimental.
180 DOI 10.1002/aic Published on behalf of the AIChE January 2012 Vol. 58, No. 1 AIChE Journal
concentration was increased to improve the scent. Three
prototypes were prepared, containing 2, 3, and 4 wt % of
linalool, respectively. Only the 4% linalool prototype showed
a satisfactory scent after the evaporation of the solvent mix-
ture. The second modication was the addition of a mild
acid, such as acetic acid (glacial, 100%, from AnalaR), to
correct the pH. Four prototypes were prepared with 0.05,
0.3, 0.5, and 1 wt % of acetic acid, respectively. A concen-
tration of 0.05% brought the pH value to 5.5, which is
exactly the skin pH.
Task S3-1, iteration 2: Simple/fast tests. The new product
formula (second iteration) is given in Table 4. Experiment 1
did not need to be performed again.
Experiments 24: These tests were successful: after the
addition of acetic acid, no phase separations were observed.
Experiment 5: The properties (density, viscosity, and
evaporation time) of the second iteration product formula
were measured and reported in Table 7. The properties did
not show drastic changes since the modications of the prod-
uct formula were quite small, and they still matched the a
priori dened constraints.
Experiment 6: The measurement of the T
90
was performed
again on the new product formula (Figure 2b). The new for-
mulation reached the 90% weight loss in a slightly longer
time than the base case formula, due to the increased amount
of linalool in the new formulation. The formula still evapo-
rated as desired.
Experiments 79: The product formula was still spray-able
and the stickiness was reduced (this is maybe due to the
lower concentration of Picaridin). The pH was still 5.5 (skin
pH).
Task S3-4, iteration 2: Difcult tests. Experiment 10: The
test was satisfactory: no phase split was observed and none
of the sensorial factors were affected by the temperature
changes.
Experiment 11: Also, this test was successful: after resting
for two months, the product formula did not show any
change.
The second iteration formula (Table 4) is the nal
product formulation which will undergo the next steps of
product development (not taken into consideration in this
work).
Design of a water-resistant sunscreen
The aim of this case study is to design a waterproof
sunscreen lotion with a sun protection factor (SPF) in the
range 1015.
The sun produces a wide range of electromagnetic radia-
tion. Ultraviolet light (UV-A, B, and C) is responsible for
sunburn and suntan and increases the risk of skin cancer.
UV-C is stopped by the ozone layer in the upper atmosphere
of the earth. Almost all of the UV-A and UV-B rays pass
through the ozone layer and cause sunburns, skin aging and
skin cancer. One of the defenses of the body against UV
radiation is the production of melanin, a pigment that results
in darkening of the skin; but this natural defense is not suf-
cient to avoid severe damages to the skin. Sunscreens are
cosmetic formulations that block UV rays.
Computer-Aided Stage (S1-D). Task S1-D1: Problem de-
nition. Consumers want a product which provides: protec-
tion from sunburns and the risk of skin cancer (which
requires protection against UV-A and UV-B) and prevention
of skin aging (main functions of the product); good material
compatibility; water-resistance; good cosmetic properties;
and sensorial factors (pleasant color and odor, pleasant skin
feeling, etc.); low price, long durability (it should not be
necessary to apply the lotion several times during the day);
low toxicity; good stability; user friendliness, such as a spray
product; long shelf life.
Except for the main product activities, a spray sunscreen
lotion and a spray insect repellent are quite similar in terms
of target properties since they are both personal care spray
products. Therefore, the translation of the above performance
criteria to the technical specications is the same as
described for the insect repellent. Table 2 gives the con-
straints and their values for the sunscreen lotion.
Task S1-D2: AI identication. Three main functions of the
product were identied in the previous task. Protection from
sunburns and prevention of skin cancer can be achieved by
providing protection from both UV-A and UV-B rays. A sin-
gle chemical that provides screening for both types of UV
radiations could not be found, therefore, two different AIs
were added. The skin aging can be prevented with antioxi-
dants. In addition, inorganic pigments like titanium dioxide
or zinc oxide are usually added to formulations, since they
are opaque to light and provide a physical barrier for radia-
tions. Since the product should be water-resistant, oil soluble
chemicals need to be selected, and the least toxic compounds
were preferred. Avobenzone (4-tert-butyl-4
/
-methoxydiben-
zoylmethane) was selected as UV-A blockers
45
; octyl salicy-
late (2-ethylhexyl 2-hydroxybenzoate) was selected as UV-B
blocker
45
; a-, b-carotene, and vitamin A were selected as
antioxidants
45
; the selected inorganic pigment was titanium
dioxide
45
(TiO
2
).
The properties (related to solubility) of the AIs (except
TiO
2
, which cannot be dissolved in the product, but dis-
persed) are shown in Table 8. The average value of the solu-
bility parameters of the AIs (d
AIs
= 20.3 MPa
1
=2
), inorganic
pigment excluded, was used to calculate the numerical val-
ues for the upper and lower bounds of the constraints on the
solubility parameter of the solvent mixture (d) and the addi-
tives (d
add
).
Task S1-D3: Solvent mixture design. The esters database
was retrieved from the database library since all the selected
AIs are all oil soluble chemicals (except TiO
2
, which is in-
soluble) and the aim of this case study is to design a water-
resistant product. In addition, esters are widely used in
Table 8. Melting Points (Experimental Data from Databases)
and Solubility Parameters of the AIs Chosen for the
Sunscreen Lotion (All Predicted with a M&G Method)
Kind of AI AI T
m
d
UV lters Avobenzone 354538 23.44
Octyl salicylate \298 21.50
Antioxidants a-Carotene 430* 17.71
b-Carotene 451452 17.92
Vitamin A 334336 20.69
Unit of measure: T
m
(K), d (MPa
1/2
).
*Predicted with a M&G method.
AIChE Journal January 2012 Vol. 58, No. 1 Published on behalf of the AIChE DOI 10.1002/aic 181
personal care and pharmaceutical applications for their
interesting functions (between which, the moisturizing
effect). The models selected for the mixture target proper-
ties were the same as those selected for the previous case
study (Appendix). The temperature considered in the design
was 300 K.
The MIXD algorithm was applied for all the property con-
straints. Results are shown in Table 9. In Mixtures 1, 2, and
3 the second compounds are structural isomers; the same for
Mixtures 4, 5, and 6. Isomers can have very similar property
values. In fact, the property values of the Mixtures 1, 2, and
3 are close to each other, as well as the property values of
the Mixtures 4, 5, and 6. In addition, methoxyacetaldehyde
is present in all the mixtures in high concentrations.
The mixture classication algorithm
1
was applied and all
the mixtures were found to be of the PNA/PNA type (esters
are polar but non associating uid, PNA). Hence, verication
with rigorous models was not necessary.
Toxicity was chosen as the preferred PI. The least toxic
mixture is Mixture 4, methoxyacetaldehyde1,1-dimethyl-
propyl 3-methylbutanoate (the highest the value of LC
50
, the
least toxic the mixture).
Task S1-D4: Additives identication. According to the per-
formance criteria (task S1-D1), the quality to enhance was
the scent. Additional qualities to enhance/augment/add for a
sunscreen lotion are the UV lters protection, the stability,
and the protection from microbial growth or undesirable
chemical changes. Chemicals that can provide these qualities
were retrieved from the databases. Aroma compounds were
selected (more details on the selection of the aroma com-
pounds can be found in previous publications
30
) to enhance
the scent: para-menth-3-yl phenylacetate or iso-propyl salicy-
late.
39
Octocrylene was selected to augment the UV lters
protection and the product stability. Parabens were chosen to
prevent the decomposition by microbial growth or by unde-
sirable chemical changes.
The above additives are all esters, as the AIs and the sol-
vents selected in previous tasks. The constraint on the solu-
bility parameter value of the additives (see Table 2) was
checked. All the additives matched this constraint, except
para-menth-3-yl phenylacetate, which was rejected. Iso-
propyl salicylate was therefore selected as the aroma com-
pound for the sunscreen lotion.
In Table 10, the details of the base case formula are
given, along with the suggested composition. The composi-
tion of the AIs in the formulation is a critical parameter
since the SPF depends not only on the type of sun blocker
compounds selected but also on their composition. The
relation between the composition of AIs and the SPF has
not been established yet. Cheng et al.
13
showed that with
an AIs concentration equal to 9.6 wt %, a SPF of 6.4 is
reached. The objective of this case study is to reach a SPF
of 1015, therefore, a total concentration of AIs equal to
20 wt % is proposed.
Table 10. Base Case: First, Second, and Third Iteration Formulations for the Sunscreen Case Study
Family Chemical
% w
i
Base Case First Iteration Formula Second Iteration Formula Third Iteration Formula
AIs Avobenzone 4.00 4.0 4.0 4.0
Octyl salicylate 4.00 4.0 4.0 4.0
a-Carotene 2.00
b-Carotene 2.00 4.0
Vitamin A 4.00 4.0 4.0 4.0
TiO
2
4.00
40%
w
ZnO dispersion 10.0 10.0 10.0
Vitamin E acetate 4.0 4.0
Solvent mixture Methoxyacetaldehyde 66.70
2,2-Dimethylpropyl butanoate 8.20
Butyl acetate 69.0 69.0 69.0
Additives Octorylene 1.70 1.7 1.7 1.7
Propyl paraben 1.70 1.7 1.7 1.7
Iso-propyl salicylate 1.70
Linalool 1.7 1.7 1.7
Almond oil 2.0
Note that in this case study, the base case formulation does not correspond to the rst iteration formulation, due to modications made during stage 2 (experi-
mental planning stage).
Table 9. Mixtures Matching the Target Properties, Their Property Values, and Stability Information
for the Sunscreen Case Study
No. Mixtures x
1
d m q LC
50
T
90
C Phase Split
1 MacAl 2,2-dimethylpropyl butanoate 0.89 18.95 0.53 832.53 3.63 1017.8 1.40 Stable
2 MacAl tert-butyl pentan3oate 0.89 18.95 0.53 832.53 3.63 1017.8 1.40 Stable
3 MacAl isobutyl isopentanoate 0.89 18.93 0.48 830.01 3.65 878.7 1.40 Stable
4 MacAl 1,1-dimethprop. 3-methbut. 0.91 18.91 0.52 825.23 3.86 846.7 1.41 Stable
5 MacAl 2,2-dimethprop. 3-methbut. 0.91 18.92 0.53 827.20 3.80 940.2 1.41 Stable
6 MacAl isobutyl 3,3-dimethbutanoate 0.91 18.92 0.53 827.20 3.80 940.2 1.41 Stable
Unit of measure: d (MPa
1/2
), m (10
6
m
2
/s), q (kg/m
3
), LC
50
(mol/m
3
), T
90
(s), Cost ($/kg).
MacAl stands for methoxyacetaldehyde and dimethprop. 3-methbut stands for dimethylpropyl 3-methylbutanoate.
182 DOI 10.1002/aic Published on behalf of the AIChE January 2012 Vol. 58, No. 1 AIChE Journal
Experimental Planning Stage (S2). The following list of
experiments was found to be necessary (it resembles the list
of the previous case study):
1. Verication of the solubility of every AI in the sol-
vent mixture.
2. Verication of solubility of every additive in the sol-
vent mixture.
3. Production of the prototype formula and verication
of its phase stability.
4. q, m measurement for pure solvents, solvent mixture,
and formulation.
5. T
90
measurement for pure solvents, solvent mixture,
and formulation.
6. Verication of spray-ability through a nozzle.
7. Verication of the sensorial factors and cosmetic
properties: appearance (turbidity/color), odor, stickiness,
greasiness, irritating power, and soothing effect.
8. Verication of the pH of the formula.
9. Verication of stability at different temperatures than
the room temperature.
10. Verication of shelf life.
11. Verication of SPF.
As in the previous case study, Experiments 18 are the
simple and fast tests whereas Experiments 911 are the dif-
cult and time consuming tests. Table 11 lists the perform-
ance criteria for the insect repellent lotion (sub-task S1-D1)
and the corresponding experiments to be performed.
The list of experiments includes also the verication of
those performance criteria that were not included during the
computer-aided design stage (Experiments 710).
Avobenzone (98 wt %) and octocrylene were obtained
from Meryer, octyl salicylate ([99 wt %) from SAFC, natu-
ral vitamin A from H
2
EI (note that according to the informa-
tion in Table 8, vitamin A should be solid; instead, vitamin
A from H
2
EI is liquid). Since a and b-carotene have very
similar properties, only b-carotene was purchased (from
Wako). Iso-propyl salicylate was not purchased since linal-
ool (from the previous case study) was available (linalool
solubility parameter matches the constraint on the solubility
parameter, it is therefore compatible with the other ingre-
dients). Nonyl paraben was not found in the market therefore
propyl paraben (from Sigma-Aldrich) was purchased.
Methoxyacetaldheyde could not be obtained; therefore,
an alternative solvent/solvent mixture needed to be
considered. All the other mixtures in Table 9 could not be
considered as alternatives, since methoxyacetaldheyde
appears in all of them. The opportunity to replace the sol-
vent mixture was investigated by going back to Stage 1
(computer-aided design) following the loop L1 in Figure 1.
At task S1-D3 (mixture design task) the constraint on the
toxicity parameter LC
50
was relaxed (LC
50
[ 0.31 mol/m
3
)
and a new simulation was performed. Butyl acetate ([99.5
wt %, Sigma-Aldrich) fullled the requirements, and it was
chosen since it was available in stock.
Because of the well known toxic properties of small size
powders, TiO
2
was replaced by an available suspension of
ZnO (ZnO is also used as physical blockers for the UV-radia-
tions in many sunscreen products). This suspension consists of
inorganic pigments dispersed in capric/caprylic trygliceride.
The solubility of the capric/caprylic trygliceride in the blend
constituted of all the other ingredients of the formulation
(TiO
2
excluded and butyl acetate instead of the solvent mix-
ture designed in task S1-D3) was checked following the loop
L1 of Figure 1 that leads from Stage 2 back to Stage 1 (task
S1-D2) of the methodology. The available property models
were employed to estimate the trygliceride solubility parame-
ter (17.72 MPa
1
=2
), which was found to be very close to the sol-
ubility parameters of the other AIs (see Table 8). Therefore, it
was decided to use this ZnO dispersion for the experimental
validation. After all these considerations, the base case for-
mula changed to the rst iteration formula (see Table 10).
The experimental set-ups employed in this case study
were the same as the insect repellent example, except for the
solubility tests and the SPF test (which was not performed
for case study 1):
Solubility test: since the sunscreen lotion involves
numerous AIs and additives, solubility tests were performed
separately for each AI and additive in the solvent mixture
(in this case pure solvent, butyl acetate) to identify which
chemicals cause miscibility problems. The AI (or additive)
concentrations in butyl acetate (w
AI
|
AIbut.ac.
) for the solubil-
ity tests were calculated as follows:
w
AI
[
AIbut:ac:
=
w
AI
[
formula
(w
AI
w
but:ac
)[
formula
(2)
w
AI
and w
but.ac.
are the concentration of AI and butyl
acetate, respectively. The subscript AIbut.ac. stands for
solution of one AI in the solvent whereas the subscript
Table 11. Experiments Employed to Verify the Performance Criteria for the Sunscreen Lotion
Performance Criteria Target Property Considered in Stage 1? Experiments Planned in Stage 2
Protection from sunburns Yes Exp. 11
Prevention of skin cancer Yes Exp. 11
Prevention of skin aging Yes
Material compatibility Yes
Water resistance Yes
Good sensorial/cosmetic factors Only odor Exps. 7 and 8
Low priced Yes
Durability T
90
Yes Exp. 5
Toxicity LC
50
Yes *
Stability d, DG
mix
, TPD Yes Exps. 1, 2, 3, and 9
Experiment 10 veries stability in a range of temperature around 300 K whereas in stage 1 only one temperature was used for the design (300 K).
AIChE Journal January 2012 Vol. 58, No. 1 Published on behalf of the AIChE DOI 10.1002/aic 183
formula means the rst iteration formula shown in Table
10. When adding a solid to a liquid it is recommended to
stir at a higher temperature (around 313323 K) than the
room temperature, to promote the dissolution process.
The SPF test follows the guidelines of FDA.
46
An arti-
cial source of light is employed: a solar simulator (Oriel
#96000, 150-W) with a total power at the exit port of 8.806
mW/m
2
(UVA: 6.906 mW/m
2
; UVB: 1.900 mW/m
2
) was
used. The test site area is the inner part of the forearm, di-
vided into ve test sub site areas of 2.5 cm diameter (each).
Each sub site area within a test site area is subjected, for a
time interval, to an articial light source for the determination
of the minimal erythematic dose (MED), for a series of time
intervals. The rest of the skin around the sub site area is
covered. At rst, the MED for the unprotected skin (US) is
measured in one test site area with the following time interval
series: 60, 75, 94, 118, and 146 s (geometric series 1.25
n
).
The time interval series for the protected skin (PS) test is
selected in this way: the MED on unprotected skin is
multiplied for the supposed SPF and this time constituted the
central time interval of the geometric series. The SPF corre-
sponds to:
SPF =
Exposure time interval MED (PS)
Exposure time interval MED(US)
(3)
Uncertainty of the test is related to the interpretation of
test results and depends on the individual perception of the
minimal erythematic dose response (the readings can vary of
20%, different reaction to UV light radiations in different
people). Since only one volunteer is to be used for the test
(for the SPF determination), the value obtained would be
just indicative.
Experimental Validation Stage (S3). All results from this
stage are summarized in Table 12.
Task S3-1, iteration 1: Simple/fast tests. Tests 18 (simple
tests) were here performed.
Experiment 1: The solubility of the AIs and additives in
butyl acetate was veried. The composition of the solutions
produced for the solubility tests (calculated as in Eq. 2) and
the test results are shown in Table 13. All experiments were
satisfactory except the solubility test of b-Carotene.
Tasks S3-2 and S3-3, iteration 1: Problem identication and
amendments. b-Carotene was found to be not soluble in
butyl acetate. Since only one AI out of ve was found to
have miscibility problems with the solvent, it was decided to
substitute it with another ingredient (instead of modifying
the solvent mixture, pure solvent in this case study). There-
fore, the loop L2 in Figure 1 for not suitable ingredients
(linking task S3-D3 with actions in Stage 1) was followed.
The antioxidants database was searched and vitamin E ace-
tate was chosen as a replacement of b-Carotene. Vitamin E
acetate is a form of powdered vitamin E that is naturally
converted by the body to vitamin E. It is an ester and its sol-
ubility parameter (16.91 MPa
1
=2
) is close enough to the solu-
bility parameters of the other AIs. Vitamin E acetate was
purchased from Opal.
The composition of the second iteration formula is shown
in Table 10, and compared with the base case and the rst
iteration product formulations.
Task S3-1, iteration 2: Simple/fast tests. Experiment 1: The
solubility test on the new AI (vitamin E acetate) was performed.
Vitamin E acetate was found to be soluble in the solvent.
Experiments 23: The solubility tests of the additives in the
solvent (Experiment 2) did not need to be performed again.
Table 12. Experimental Results for the Sunscreen Case Study: First, Second, and Third Iteration
No. Experiment
Result: First
Iteration Result: Second Iteration Result: Third Iteration
1 AIs solubility in the solvent mixture One AI does not
dissolve
Successful
2 Additives solubility in the solvent mixture Successful Successful (performed only
for almond oil)
3 Phase stability of the overall formula Successful Successful
4 m and q of pure compounds, solvent mixture,
and formula
Matching targets* Matching targets
5 T
90
of pure solvents, solvent mixture, and formula Satisfactory Satisfactory
6 Formula sprayability Successful Satisfactory
7 Appearance (turbidity/color), odor, stickiness,
greasiness, irritation, soothing effect
Odor not satisfactory
,
soothing effect could be
improved
Odor not satisfactory
,
soothing effect improved
8 pH Satisfactory Satisfactory
9 Stability at different temperatures than 300 K (278
and 318 K)
Not satisfactory
10 Shelf life Satisfactory with condition
The odor is not satisfactory but this factor is taken into consideration for improvements because of chemicals availability issues.