9th International Conference LCA of Food San Francisco, USA 8-10 October 2014

Specialty Food Ingredients – Environmental Impacts and Opportunities

ABSTRACT
Specialty food ingredients (SFIs or just ‘specialty ingredients’) such as emulsifiers, food enzymes, hydrocolloids and cultures are typical-
ly applied in small amounts (<1% w/w of the final food product). Generally, little attention has been given to specialty ingredients in the
LCA community (Foster et al 2006, BCFN 2011). DuPont Nutrition & Health (N&H) is a leading producer of specialty food ingredients
worldwide. During the last five years, DuPont N&H has completed cradle-to-gate LCAs of a wide range of specialty ingredients, based
on both attributional and consequential modeling. The average carbon footprint of DuPont N&H products is 3-4 kg CO2e per kg (cradle-
to-gate) but they only represent a small share of the final consumer products carbon footprint. More importantly, most solutions enable
significant reductions of our customer’s footprint by enabling replacement of animal derived raw materials (with e.g. soy protein), in-
creasing processing efficiency or enabling reduced food waste in retail and households. DuPont N&H has identified more than 70 ‘sus-
tainable solutions’ and LCA screenings of 40 cases show that they can help customers reducing between 10 and 100 kg CO2e per kg spe-
cialty ingredient applied.

Keywords: Specialty Food Ingredients, Life Cycle Assessment, Carbon Footprint, Food Waste, Sustainable Food

1. Introduction

A significant number of food companies apply life cycle assessment (LCA) to quantify the environmental
burdens associated with their products. LCA studies have been published on a wide range of food, including
dairy products, meat products, vegetables, fruits, bread, alcoholic and non-alcoholic beverages, confectionary etc.
(Foster et al 2006, BCFN 2011). Few studies, however, are available on specialty ingredients such as:
 Emulsifiers (e.g. used to strengthen and soften the dough in bread)
 Food enzymes (e.g. bakery enzymes used to extend periods for which breads stay fresh)1
 Hydrocolloids (e.g. pectin used as gelling agent in jam and marmalade)
 Bacteria cultures (e.g. acidifying cultures applied in the production of yoghurt and fresh fermented milks)
 Antimicrobials and antioxidants (e.g. rosemary extract used to prevent pathogens and oxidation in meat)
 Reduced calorie sweeteners (e.g. xylitol used as sweetener in chewing gum).
 Soy protein (e.g. soy protein isolate used to replace animal protein in different food applications)

Apart from a few examples such as soy proteins and some sweeteners, specialty ingredients are typically used
in small amounts (<1% w/w) to obtain certain functions such as extended shelf-life/freshness or improved taste,
texture and mouth feel. In most existing LCAs of food products, impacts associated with SFIs are only modeled
coarsely. But it is also important for food producers to understand the environmental impacts associated with
SFIs, and more importantly to be aware of the opportunities they bring in terms of replacement of high impact
raw materials, improvements in processing efficiency and not least reduction of food waste.
As a leading producer of SFIs, DuPont N&H decided three years ago to carry out cradle-to-gate LCAs on all
main product categories. This paper shows examples of results from these studies as well as results from LCA
screenings of the use stage. In the discussion part of the paper, attention will be given to methodological chal-
lenges.

1
Several scientific LCA studies on enzymes including food enzymes are available in the literature (Jegannathan and Niel-
sen 2013).
9th International Conference LCA of Food San Francisco, USA 8-10 October 2014

2. Methods

In terms of cradle-to-gate LCA studies there has been completed six studies which have been third party re-
viewed in accordance with the ISO 14044 standard (two additional studies are in review). Our goal is that all our
main product categories should be analyzed through 3rd party reviewed ISO 14044 compliant LCA studies by
2015. In addition, 40 LCA screenings of use stage impacts have been completed for a range of products and ap-
plications that we call sustainable solutions. The following sections address key methodological choices related
to our cradle-to-gate LCAs as well as the LCA screenings of the use stage.

2.1. Goal and Scope

Our cradle to gate (or cradle-to-customer) LCAs include all life cycle stages from raw material acquisition to
final dispatch of products to our customers and includes in many cases even outbound transport to the average
customer. The functional unit can vary from study to study but reflects a reference flow of 1 kg specialty ingre-
dient in all studies. Specialty ingredients are sold because of their function which is the reason why we often de-
scribe them as ‘solutions’ rather than ‘products’. The functionality is therefore always described in detail, and
relevant adjustments in the reference flow are made when comparisons are made between different SFIs.

2.2. Life Cycle Inventory (LCI)

In DuPont N&H consequential and attributional life cycle inventory modeling are seen as complementary ra-
ther than competing which will be explained later. Both approaches are therefore included in our detailed third
party reviewed LCA studies as well as LCA screenings of the use stage.
In consequential modeling co-product allocation is consistently avoided by system expansion (substitution)
and constrained processes are excluded from the analyzed product system. As an example, this means that hy-
dropower would not be included in the grid mix for electricity in a country where all hydropower is fully utilized
(e.g. Norway). Modeling of indirect land use change (ILUC) is only included in sensitivity scenarios.
Attributional modeling represents an approach where co-product allocation is typically handled with alloca-
tion, and where no distinctions are made between constrained and unconstrained processes (Sonnemann and Vi-
gon 2011 p74). In DuPont N&H economical allocation is generally used for handling of co-product allocation in
our attributional models, which is consistent with the ecoinvent V3 default allocation model (Weidema et al
2013).
Background processes have generally been modeled based on data from the ecoinvent V2.2 database (in all
studies completed before 2014), but other data have been used when these have been deemed to represent a
higher level of data quality. This includes, for example, data on country-specific marginal and average electricity
which has been modeled separately in collaboration with 2.-0 LCA consultants (Schmidt et al. 2011). Ongoing
studies are based on Ecoinvent 3 which allows for more consistent modeling, where either economical allocation
or system expansion is used in all background processes.

2.3. Life Cycle Impact Assessment (LCIA)

The Recipe midpoint model has been used as the default method for impact assessment (in all studies com-
pleted before 2014) but other LCIA methods, including end-point methods, have been applied for sensitivity
studies. Ongoing LCAs applies the ILCD LCIA method available in SimaPro 8 which is applied as it is recom-
mended by the European Product Environmental Footprint (PEF) Guide and is a result of a consensus project
aimed at establishing current best practice for LCIA (European Commission 2012).

2.4. Use stage modeling

In most cases, one type of specialty ingredient can typically be used in many different food products where it
can have various functions. Furthermore, specialty ingredients are often used in various combinations and in dif-
ferent parts of the world. Needless to say, it is complex to model the use stage without focusing on concrete cas-
es. The use stage has therefore been modeled separately and at screening level, focusing on specific applications
9th International Conference LCA of Food San Francisco, USA 8-10 October 2014

in different parts of the world. Three scenarios are calculated for each solutions which represent attributional
modeling, consequential modeling (incl ILUC) and consequential modeling (excl ILUC) which is chosen as our
default model. The advantage of the different models is that it allows us to adapt to specific customer prefer-
ences. In terms of LCIA the screenings only address the carbon footprint based on characterization factors in the
ILCD LCIA method in Simapro 8.
Conceptually, it is possible to distinguish three types of SFIs based on their functions: Specialty ingredients
that mainly are applied to enable a substitution of raw materials (or modify recipes), ingredients that mainly aim
at increasing processing efficiency and finally ingredients that serves to prolong the freshness or shelf-life of
food products, see Figure 1.
DuPont N&H
Specialty Food Ingredients

Replacement of high Improvement of Reduction of

As illustrated the distinguishing feature between the three groups of SFIs depend on which part of the food
value chain they mainly address:

 Upstream SFIs: Some SFIs address upstream impacts by enabling modifications in raw material compo-
sition e.g. by replacing animal protein and fat with vegetable protein and fat (Figure 1 left). This applica-
tion can be divided in two sub-categories: direct upstream where the SFI itself constitutes the replace-
ment (e.g. soy protein that replaces milk protein) and indirect upstream where the SFI has an indirect
role allowing the replacement to take place (e.g. hydrocolloids that facilitate a reduction of animal fat or
a replacement with vegetable fat in dairy, while maintaining similar taste and texture).
 Processing SFIs: Another group of SFIs addresses the food processing directly – e.g. by allowing for
larger throughput and/or increased efficiency (Figure 1 center).
 Downstream SFIs: These are SFIs that address the food products’ freshness and their taste and texture as
perceived by the end customers in a typical use scenario (Figure 1 right). This application has the poten-
tial to reduce food waste significantly, especially for categories of food products where the dominating
reason for food waste is that it has passed a date label, has gone moldy or rotten, looked, smelt or tasted
bad.

All three functions have a significant potential to promote sustainability in the food value chain – especially
upstream and downstream SFIs that address the most important hot-spots of average food products (raw materi-
als and food waste). A large generic group of SFIs such as food enzymes or hydrocolloids can have several of
the above mentioned functionalities, while more specific products typically have a more narrow functionality re-
lated to one of the functions.
9th International Conference LCA of Food San Francisco, USA 8-10 October 2014

3. Results

3.1. Example of comparative LCA on Xylitol/Xivia®

One of the first LCAs completed by DuPont N&H was a comparative LCA on the sweetener xylitol, which
was reviewed by a third party panel in 2011. Two production methods were compared – one based on a side
stream from the pulp and paper industry (the wood based concept used by DuPont), and one based on corn cobs
(used by most of our competitors). The study was completed by Earthshift consultants and applies attributional
modeling but with the use of system expansion (substitution) for energy and waste related foreground processes.
In this particular study, Impact 2002+ has been used for life cycle impact assessment (LCIA). A significant dif-
ference was anticipated, but it was a surprise to most that the wood based technology actually generated 85-99%
lower impacts across all investigated impact categories (see Figure 2).

Impact categories Xylitol (corn cob based) Xivia® (wood based)

Terrestrial ecotoxicity
Terrestrial acid/nutri
Respiratory organics
Respiratory inorganics
Ozone layer depletion
Non-renewable energy
Non-carcinogens
Mineral extraction
Land occupation
Ionizing radiation
Global warming
Carcinogens
Aquatic eutrophication
Aquatic ecotoxicity
Aquatic acidification
Percent
0 10 20 30 40 50 60 70 80 90 100
Figure 2. Characterized results from comparative LCA (based on Impact 2002+) of corn cob based xylitol and
DuPont’s wood based xylitol (Xivia®) based on Dahliwal, Hamilton and Laurin (2010).

In terms of global warming potential the wood based technology contributed 3.6 kg CO 2e per kg Xivia® (cra-
dle-to-gate) while the corn cob based technology contributed 38.6 kg CO2e per kg xylitol from cradle-to-gate
(Dahliwal, Hamilton and Laurin 2010). Results also published in Danisco (2011) and Dahliwal and Thrane
(2011).

3.2. Carbon footprint and hot-spots for different groups of SFIs

Based on findings from LCA studies of SFIs as well as DuPont’s Corporate Footprint based on WRI &
WBCSD (2004 and 2010), it can be concluded that average carbon footprint of DuPont N&H’s specialty ingre-
dients is in the range of 3 to 4 kg CO2e per kg from cradle-to-gate, excluding ILUC. The main contribution
comes from raw materials (53%), while processing represents 35% of the cradle-to-customer footprint. Inbound
and outbound transport both represents 6% respectively (Thrane 2011, Dalgaard et al. 2013).
With a typical application level of less than 1% w/w of the customer’s final product, this suggests that spe-
cialty ingredients in most cases have a modest contribution to a food product’s total footprint. There are cases,
however, where SFIs are used in larger amounts such as for soy proteins, and in the case of Xivia® in chewing
gum applications.
9th International Conference LCA of Food San Francisco, USA 8-10 October 2014

Emulsifiers: For average emulsifiers, the raw materials (mainly vegetable oils) constitute about 80% of the
cradle-to-gate footprint, and the average emulsifier has a carbon footprint of 3-5 kg CO2e per kg product from
cradle-to-gate (Muñoz 2014). Both attributional and consequential modeling has been applied, but the results are
within the range mentioned above in both LCI models.
Food enzymes: LCA of enzymes have been conducted for several years in DuPont Industrial Biosciences. In
terms of carbon footprint, the most important life cycle stage for food enzymes are processing (fermentation,
formulation and recovery) followed by raw materials use in the fermentation process. The carbon footprint rang-
es between 1 and 10 kg CO2e per kg enzyme from cradle-to-gate mainly depending on the type and quantity of
raw materials, the fermentation time and the concentration (Dettore 2014). Again both consequential and attribu-
tional modeling has been used and both approaches are represented within the indicated carbon footprint range.
Hydrocolloids: In terms of carbon footprint, the processing stage is important for most hydrocolloids. Pro-
cessing aids also have a noteworthy contribution to the footprint, but it should be stressed that hydrocolloids are
a very diverse group of ingredients with large variations in hot-spots and impacts. The typical carbon footprint
ranges from 1-12 kg CO2e per kg hydrocolloid from cradle-to-gate. This range covers many different hydrocol-
loids of which most have been analyzed through screening level LCAs apart from a third party reviewed LCA of
pectin published in 2011 (Thrane 2011). Consequential modeling has been applied in combination with attribu-
tion modeling in the third party reviewed LCA which showed only a minor difference in impacts.
Cultures: Similar to most hydrocolloids, the processing stage represents an important contribution to the car-
bon footprint for cultures – mainly due to the energy use. Two types of direct inoculation cultures exist: Frozen
cultures and freeze dried cultures, which are concentrated and therefore more efficient. Adjusted for functionali-
ty, the cradle-to-gate footprint for both types of cultures are in the range between 8 and 12 kg CO2e per kg. Cul-
tures are added at rather small dosages in yoghurt and cheese; between 0.5 and 2 kg frozen cultures per 10,000
liters of milk. As for most SFIs their contribution to the final products footprint is therefore small. In a cradle-to-
customer perspective, the carbon footprint for outbound transport can be important, especially for frozen cul-
tures. The reason is that frozen cultures have to be stored at minus 50°C to 60°C and frequently require air
transport – especially for transcontinental shipments. In cases where outbound transport involves airfreight,
freeze dried cultures are likely to offer a low carbon footprint alternative to frozen cultures. Both consequential
and attributional modeling has been applied and again no significant differences in impacts have been identified
apart from what is covered by the suggested interval (Thrane 2012, Thrane 2013a, Thrane 2013b).
Reduced calorie sweeteners: For this category, only Xivia® has been analyzed and the study shows that the
raw material- and processing stage represent the largest carbon footprint (due to energy use). The carbon foot-
print of Xivia® is within the same range as our average product, 3-4 kg CO2e per kg (Dahliwal, Hamilton and
Laurin 2010). But as Xivia® is the only sweetener type that has been analyzed, it is not possible to conclude that
this product is representative for the other products in this category.
Protein solutions: For soy protein concentrate (min 65% protein per dry matter) and soy protein isolate (min
90% protein per dry matter) the main contribution comes from the processing stage due to energy use. A detailed
LCA is being completed but final results are not yet available. Preliminary results indicate that the carbon foot-
print is around 2-4 kg CO2e per kg protein product. Consequential modeling suggests a lower impact compared
to attributional modeling.
Apart from the product groups mentioned above, DuPont N&H also produces antimicrobials and antiox-
idants (covering products such as Natamax®, Nisaplin® and GUARDIAN® Rosemary Extract) as well as a range
of other products that could be termed ‘other’ which include fibers, rare sugars, soy lecithin etc. Considering the
large number of ingredients and applications, the scope has been limited to cover the above mentioned groups in
the present article.

3.3. LCA screenings of avoided burdens (use stage impacts)

As mentioned, the use stage has also been modeled by LCA screenings focusing on the carbon footprint. So
far, more than 70 solutions have been identified and 40 of these have been quantified by LCA screenings. In a
typical case, 1 kg specialty ingredient contributes to avoiding 10-100 kg CO2e (net) compared to a situation
where the ingredient is not used. All LCA screenings have been made according to both attributional and conse-
quential modeling, but the results presented in the following are only based on consequential modeling, which in
most cases represent a more conservative estimate of potential savings.
9th International Conference LCA of Food San Francisco, USA 8-10 October 2014

Direct upstream SFIs: One example of direct raw material replacement is the use of Xivia® instead of corn
cob based xylitol in chewing gum. The net avoided impact per kg applied Xivia® is the difference in impact be-
tween one kg corn cob based xylitol and one kg Xivia®, which amounts to 35 kg CO2e. LCA screenings shows
that the carbon footprint is reduced by 42-85% per kg chewing gum when switching from corn cob based xylitol
to Xivia®. The large interval represents differences in xylitol content in the final recipe with all ingredients in-
cluding gum base. The modeled xylitol content ranges from 5% to 55% (Dalgaard et al. 2013a). Another exam-
ple could be the use of soy protein to replace milk or meat protein in different applications such as dairy prod-
ucts, sausages and burgers. Updated third party reviewed LCA results of soy protein isolate are not yet available,
but our LCA screenings suggest that 1 kg isolated soy protein contributes to reducing 10 kg CO2e (net) in appli-
cation where it replace milk protein2 and even more when replacing meat protein. An example of a beef burger,
where the beef content is partly replaced with textured soy concentrate, shows that the carbon footprint of the
beef can be reduced with as much as 35% (Dalgaard et al. 2013 a and b).
Indirect upstream SFIs: An example where the specialty ingredient has an indirect effect on raw materials
could be a pectin based ingredient solution, such as GRINDSTED® SB555, that makes it possible to replace milk
protein with vegetable protein in yoghurt, without compromising organoleptic properties. An LCA screening
shows that 1 kg GRINDSTED® SB555 contributes to reducing 12 kg CO2e per kg (net) in a specific application
where it allows for using vegetable protein instead of milk protein in yoghurt. The solution allows the footprint
of the final yoghurt to be reduced with 18%. Other solutions exist which enable replacement of meat protein in
sausages, burgers etc. (Dalgaard et al. 2013b). Another example of indirect upstream substitution could be the
enzyme Alphalase® AP4 applied in beer production. This solution makes it possible to brew a good quality beer
where 60% of the malt is replaced with barley. LCA screenings show that 1 kg Alphalase® AP4 contributes to
reducing 100 kg CO2e (net) in this application and the footprint of the final beer is reduced by 12% (Dalgaard et
al. 2013a).
Processing SFIs: An example of processing SFIs is FoodPro® Cleanline - an enzyme solution which can sig-
nificantly reduce the impacts of ultra high temperature processing of milk in dairies (UHT). This solution pre-
vents fouling in the UHT processing systems, leading to higher capacity while reducing the need for cleaning
cycles that otherwise would require significant amounts of water, energy and cleaning chemicals. An LCA
screening shows that 1 kg FoodPro® Cleanline contributes to reducing 35 kg CO2e per kg (net) when applied in
UHT processing. In a large scale dairy with 6 UHT lines, this translates in to annual net savings of 322 tons
CO2e (Dalgaard et al. 2013b).
Downstream SFIs: Finally an example of a solution that addresses downstream food waste is the use of emul-
sifier and enzyme blends in bakery products. Food waste is a major issue, and bakery products are one of the
worst culprits for avoidable food waste. Approximately 30% of bakery goods like bread, cakes and cookies are
wasted in the households and according to WRAP about 90% of this waste occurs because the products are not
perceived as fresh (Gustavsson et al. 2011, Quested and Johnson 2009). DuPont has developed enzyme solutions
that apart from providing the usual bread softness and crumb elasticity, also keep the bread moist longer. The
enzymes are sold under the Powerfresh® brand name and are often used in combination with emulsifiers. An
LCA screening has been completed for a use of 50% higher dosage of Powerfresh® (increase from 20 gram to 30
gram enzyme dosage per 100 kg flour) in toast bread in North America, enabling the bread to stay fresh for 21
days compared to 14 days with the lower dosage. Based on a model of the relationship between freshness and
bread waste, calibrated with data from Quested (2013), it is estimated that the bread waste in retail and house-
holds is reduced by 30%. The LCA screening shows that the carbon footprint is reduced with 12% per kg bread
consumed in this case – and several thousand kg CO2e is avoided per kg extra enzyme that is applied. This ex-
ample involves our newest and most sophisticated enzymes, but even standard enzymes (and emulsifiers) can
make a huge difference in markets where enzymes are more rarely used, such as parts of Eastern Europe and
Russia.
A new exiting product range with a significant potential for food waste reduction is protective cultures such
as HOLDBAC® cultures used for spoilage and pathogen prevention in dairy and meat applications. LCA screen-

2
In this example the milk protein is modeled based on the LCAfood database where the process ''Milk powder, no quotas' has been
modified and updated based on LCI data on milk from Schmidt and Dalgaard (2012) & Dalgaard and Schmidt (2012) using the conse-
quential mode. If attributional modeling had been used instead the avoided burden would have been 15 kg CO2e per kg soy protein that
replaces milk protein (Dalgaard et al. 2013 b).
9th International Conference LCA of Food San Francisco, USA 8-10 October 2014

ings of protective cultures applied to fresh cheese shows that the carbon footprint can be reduced with nearly 5%
per kg consumed cheese – and 220 kg CO2e is avoided per kg HOLDBAC® cultures applied (Dalgaard et al.
2013b).

4. Discussion

4.1. Difference between attributional and consequential modeling

In LCAs of specialty ingredients, the difference between attributional and consequential modeling is often
modest. But there are situations where the differences actually matter. These can be situations where a part of the
product system that is constrained, is left out in the consequential model. An illustrative example is SFIs that en-
able a partial replacement of milk fat with palm oil in cream cheese. This results in a significant improvement
when attributional modeling is applied, but has no effect when consequential modeling is used. The latter pre-
sumes, according to the consequential model we have used, that milk fat is a dependent co-product constrained
by the main product milk. Hence, instead of influencing the production of milk, a change in demand for milk fat
would affect the marginal oil on the world market (namely palm oil). In terms of modeling, this implies that we
replace palm oil with palm oil – hence no effect.

4.2. Consequential modeling (pros and cons)

The advantage of consequential modeling is arguably that it better reflects the functioning of the market and
attempts to model consequences of concrete decisions or choices in a concrete market context. From a consumer
perspective, it could be the environmental impacts of buying product A versus not buying it – or buying product
A instead of product B. Both situations imply changes – which is exactly what consequential LCA attempts to
model. A hot-spot assessment is not necessarily about changes – but as soon as it is used to guide decisions or to
make strategic choices it does concern changes. Consequential modeling is tailor-made to decision support and is
better able to address the bigger picture including the direct and indirect market effects.
Consequential modeling avoids co-product allocation – often by system expansion. This leaves less room for
arbitrary choices of allocation methods, but on the other hand consequential modeling opens up for different
choices and assumption related to identification of marginal products and other market aspects. It could be ar-
gued that consequential modeling, by including markets aspects, requires too much from the LCA practitioner
and makes LCA a too broad research discipline. Some would probably argue that ‘the best is the enemy of the
good’ in this context. Ultimately one of the fundamental challenges of consequential LCA is that it can be hard
to comprehend, not least for non-experts. As an example it can be difficult to communicate that an LCA of soy
oil based emulsifier does not include soy oil in the inventory, because it is constrained by the demand for soy
meal. The consumer, customer or other stakeholder might get confused – and skeptical when they are told that
palm oil is included instead because this is the marginal oil on the world market. But this is rather a communica-
tion issue than it is a real disadvantage of consequential LCA. Another challenge is that it, by putting attention to
the marginal, could lead to a missed opportunity to collaborate with the suppliers of soy oil to reduce upstream
impacts. This does not have to be the case, but could be the outcome when results are interpreted by non experts.

4.3. Attributional modeling (pros and cons)

Attributional modeling is often described as a more normative approach with fewer considerations about the
functioning of the market. Instead it follows traditional supply chain logic and it better reflects how current sup-
ply chains are physical connected. The fewer number of variables related to the market, also reduce to option for
modeling related differences when comparing results from LCA studies. Attributional modeling is arguably easi-
er understand, and from a supply chain logic, it makes intuitively more sense that soy beans are included in an
LCA of soy oil, instead of the marginal vegetable oil (palm oil) as suggested in consequential modeling. It can
be argued that, attributional modeling better reflects the way many consumers and NGOs attribute a responsibil-
ity on the suppliers in the actual supply chain. An example is when stakeholders avoid agricultural products than
are produced in regions where deforestation is taking place – by sourcing from other regions may just reflect a
shift of burdens via indirect land use changes, and eventually the effect on deforestation may be the same.
9th International Conference LCA of Food San Francisco, USA 8-10 October 2014

The main disadvantage of the attributional approach is that it includes constrained suppliers, i.e. the results of
an LCA include emissions from suppliers that will not change their production as a consequence of a change in
the demand for the studied product. Further, the attributional approach often involves different allocation meth-
ods for multiple product output activities. Apart from making studies more difficult to compare, this approach
ignores the real effects of by-products and the allocated processes fail in some cases to maintain basic balances
such as mass balance, carbon balance, energy balance etc.
It is out of scope in the present article to make a more elaborate description of pros and cons, but it should be
clear that depending on the purpose of the LCA, both approaches have advantages and disadvantages and that it
can be fruitful to provide both perspectives in LCA studies.

4.4. Other aspects of Sustainability

The present article has focused on environmental impacts, mainly the carbon footprint. But there are obvious-
ly many other elements of sustainability such as food security, health and nutrition aspects.
Several initiatives are taking place to broaden aspects of sustainability within our own supply chain. Emulsi-
fiers and hydrocolloids can enable reductions of salt and fat (including trans-fat) in food products. Reduced calo-
rie sweeteners can obviously reduce sugar levels, and products such as Litesse® can increase fiber contents of
foods. Other examples are probiotic cultures, which can promote a healthy bacteria flora in the digestive system.
Social sustainability aspects related to food are however complex to describe and include in a short article about
SFIs, and it is highly context and culture dependent. In Mexico for example, milk competes on price with soda.
This means that solutions to make milk more affordable is likely to have a positive effect on both food security
and health – despite reducing the amount of milk solids on a per liter basis. The theme about what defines a sus-
tainable and healthy diet has recently been described in a discussion paper by Garnett (2014).

5. Conclusion

DuPont N&H continuously seeks to reduce the environmental impacts from all life cycle stages including raw
materials, transport, processing and use. It should be recognized, however, that environmental burdens associat-
ed with SFIs typically are small or insignificant in the final consumer products. This is due to the small amounts
used (typically < 1% w/w) and the relative modest carbon footprint of 3-4 kg CO2e per kg average SFI. The lat-
ter is close to the footprint of the products in which they are typically used.
More importantly, SFIs can make a significant difference in terms of reducing the footprint of the food prod-
ucts in which they are used. SFIs have the largest potential to leverage sustainability in the use stage where they
can affect all stages of the customer’s value chain. Three groups of SFIs can be distinguished depending on how
they influence the customers’ value chains:
 SFIs that influence the first stage of the customers value chain (raw materials) – e.g. by allowing re-
placement of animal based with vegetable based protein or fat.
 SFIs that contributes to improve customers manufacturing efficiency.
 SFIs that mainly address the last stages of the customer’s value chain, by enabling food products to stay
fresh longer and thereby reducing food waste.
Food enzymes are an example of specialty ingredients that can address raw material substitution, processing
efficiency and downstream food waste. Other groups of SFIs are typically less comprehensive in their function-
ality, but they can have the same potential to reduce impacts such as isolated soy protein replacing animal pro-
tein.
DuPont N&H has identified 70 solutions that allow for carbon footprint reductions in the customers value
chain. LCA screenings of 40 solutions reveal that these solutions allow for reductions between 10 and 100 kg
CO2e per kg SFI applied. But there are also examples where 1 kg specialty ingredient can lead to reductions of
more than 1000 kg CO2e by reducing food waste.
Detailed third party reviewed LCAs have been applied for cradle-to-gate LCAs, while the use stage has been
modeled through screening level LCA. Both attributional and consequential modeling is applied in all studies
and generally results vary little. There are cases, however, where significant differences emerge, especially for
dairy products. While these differences may appear inconvenient they show that methodological uncertainty can
be significant in certain cases.
9th International Conference LCA of Food San Francisco, USA 8-10 October 2014

6. References

BCFN (2011) Double Pyramid: healthy food for people, sustainable food for the planet. Barilla Center for Food
& Nutrition, Parma, Italy.
Bundgaard AM, Dalgaard R, Gilbet C, Thrane M (2014). Assessment of the potential of digestibility-improving
enzymes to reduce greenhouse gas emissions from broiler production. Journal of Cleaner Production, ISSN
0959-6526 2014.
Dahliwal H and Thrane M (2011). Applying system expansion ot handle Multi-funcionality in Xylitol Produc-
tion. ACLCA conference October 2011. http://lcacenter.org/lcaxi/abstracts/476.htm
Dalgaard R and Schmidt J H (2012). National and farm level carbon footprint of milk - Life cycle inventory for
Danish and Swedish milk 2005 at farm gate. Arla Foods, Aarhus, Denmark
Dalgaard R, Brandão M, Schmidt JH, Thrane M (2013). Corporate Carbon Footprint CY 2011. DuPont Nutrition
& Health. The report is confidential.
Dalgaard R, Thrane M, Fairs I and Schmidt JH (2013a). Sustainable solutions in Other applications. Quantifica-
tion of induced and avoided impacts. DuPont Nutrition & Health and 2.- LCA Consultants, Brabrand Den-
mark. The report is confidential.
Dalgaard R, Thrane M, Fairs I and Schmidt JH (2013b). Sustainable solutions in Dairy applications. Quantifica-
tion of induced and avoided impacts. DuPont Nutrition & Health and 2-.0 LCA Consultants, Brabrand Den-
mark. The report is confidential.
Danisco (2011). XiviaTM White Paper – Sustainable and Substantiated. Danisco A/S.
http://www.danisco.com/fileadmin/user_upload/danisco/documents/products/2e_XIVIA_White_Paper.pdf
Dettore C (2014). Personal communication with LCA expert Christopher Dettore, Product Stewardship and Sus-
tainability in DuPont Industrial Biosciences, Palo Alto CA, US.
Dhaliwal H, Hamilton M, Laurin L (2010). Assessing Sustainability of Xylitol Production – LCA of Xylitol pro-
duction using a pulp and paper waste stream in comparison with corn cobs as feedstock. Third party reviewed
by SIK, Pré Consultnat and 2-.0 LCA Consultants. Prepared for Danisco by Earthshift LLC, U.S. The report
is confidential.
European Commission (2012). Product Environmental Footprint (PEF) Guide. Deliverable 2 and 4A of the Ad-
ministrative Arrangement between DG Environment and the Joint Research Centre No N
070307/2009/552517, including Amendment No 1 from December 2010. Ispra, Italy.
Foster C, Green K, Bleda M, Dewick P, Evans B, Flynn A, Mylan J (2006) Environmental Impacts of Food Pro-
duction and Consumption: A report to the Department for Environment, Food and Rural affairs. Manchester
Business School. Defra, London.
Garnett T (2014). What is a sustainable diet? – A discussion paper. Food Climate Research Network (FCRN).
http://www.fcrn.org.uk/fcrn/publications/fcrn-discussion-paper-what-sustainable-healthy-diet Apr
Gustavsson J, Cederberg C, Sonesson U, Otterdijk O and Meybeck A (2011) Global Food Losses and Food
Waste, Food and Agriculture Organization of the United Nations (FAO), Rome.
Jegannathan RK, Nielsen PH (2013). Environmental assessment of enzyme use in industrial production - a litera-
ture review. Journal of Cleaner Production 43 (2013) 228-240. Muñoz I (2014). Life Cycle Assessment of a
series of DuPont™ Danisco® emulsifiers. DuPont Nutrition & Health and 2-.0 LCA Consultants, Brabrand
Denmark. In review. The report is confidential.
Schmidt J H, Merciai S, Thrane M and Dalgaard R (2011): Inventory of country specific electricity in LCA –
Consequential and attributional scenarios. Methodology report v2. 2.-0 LCA consultants, Aalborg.
http://www.lca-net.com/projects/electricity_in_lca/
Schmidt J H and Dalgaard R (2012). National and farm level carbon footprint of milk - Methodology and results
for Danish and Swedish milk 2005 at farm gate. Arla Foods, Aarhus, Denmark
Simone Manfredi, Karen Allacker, Kirana, Chomkhamsri, Nathan Pelletier, Danielle Maia de Souza (2012):
Product Environmental Footprint (PEF) Guide. European Comission, Joint Research Centre, Ispra, Italy.
Sonnemannn G and Vigon B (eds.) (2011). UNEP/SETAC Life Cycle Initiative workshop report: Global Guid-
ance Principles for Life Cycle Assessment Databases, United Nations Environmental Programme (UNEP),
France.
9th International Conference LCA of Food San Francisco, USA 8-10 October 2014

Thrane M (2011). Life Cycle Assessment of Pectin from Danisco A/S – cradle-to-gate LCA including distribu-
tion to customers. Third party revived by SIK. Danisco A/S, Brabrand, Denmark. The report is confidential.
Thrane M (2011). Corporate Carbon Footprint FY 09/10. Third party reviewed by Deloitte Copenhagen. DuPont
Nutrition & Health. The report is confidential.
Thrane M (2012). Life Cycle Assessment of Frozen Cultures. Third party revived by SIK. DuPont Nutrition and
Health, Brabrand, Denmark. The study is confidential.
Thrane M (2013a). Addendum to Thrane (2012). The study is confidential.
Thrane M (2013b). Life Cycle Assessment of Freeze dried cultures. Third party revived by SIK. DuPont Nutri-
tion and Health, Brabrand, Denmark. The study is confidential.
Weidema B P, Bauer C, Hischier R, Mutel C, Nemecek T, Reinhard J, Vadenbo C O, Wernet G. (2013) Over-
view and methodology. Data quality guideline for the ecoinvent database version 3. Ecoinvent Report 1(v3).
The ecoinvent Centre, St. Gallen.
WRI and WBCSD (2004). The Grenhouse Protocol – A corporate Accounting and Reporting Standard. Revised
edition. World Resources Institute (WRI) and World Business Council for Sustainable Development
(WBCSD).
WRI and WBCSD (2010). Scope 3 Accounting and Reporting Standard. 2nd draft November 2010. World Re-
sources Institute (WRI) and World Business Council for Sustainable Development (WBCSD).
Quested T and Johnson H (2009) Household Food and Drink Waste in the UK. Final report. WRAP, UK.
Quested T (2013) The Milk Model: Simulating Food Waste in the Home, Final report. WRAP UK.
http://www.wrap.org.uk/sites/files/wrap/Milk%20Model%20report.pdf