Ozone: Science & Engineering

The Journal of the International Ozone Association

ISSN: 0191-9512 (Print) 1547-6545 (Online) Journal homepage: https://www.tandfonline.com/loi/bose20

Toward Efficient Low-Temperature Ozone Gas

Sterilization of Medical Devices

Sandy A. Thill & Marc Spaltenstein

To cite this article: Sandy A. Thill & Marc Spaltenstein (2019): Toward Efficient Low-
Temperature Ozone Gas Sterilization of Medical Devices, Ozone: Science & Engineering, DOI:
10.1080/01919512.2019.1704217

To link to this article: https://doi.org/10.1080/01919512.2019.1704217

Published online: 29 Dec 2019.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=bose20
OZONE: SCIENCE & ENGINEERING
https://doi.org/10.1080/01919512.2019.1704217

Toward Efficient Low-Temperature Ozone Gas Sterilization of Medical Devices

Sandy A. Thill and Marc Spaltenstein
SteriLux SA, Prilly 1008, Switzerland

ABSTRACT ARTICLE HISTORY

An ozone generation system using V-UV excimer lamps (172nm) and an optical measurement system Received 23 August 2019
based on the Beer–Lambert law were developed for an efficient low-temperature (15 °C-30 °C) steriliza- Accepted 9 December 2019
tion of medical devices. Ozone concentrations between 100 and 1ʹ000 ppm (0.2–2 g/Nm3) were KEYWORDS
generated and the impact of several variables on the sterilization outcome were studied. The survivor Ozone; sterilization;
curves of the sterilization process were obtained for biological indicators (BIs) inoculated with heat-sensitive medical
Geobacillus stearothermophilus spores. devices; optical ozone
A lag phase was observed before the start of the actual sterilization which followed first-order measurement system; V-UV
log-linear kinetics. The results demonstrated the feasibility to sterilize medical devices with ozone radiation to generate ozone
at ambient temperatures, reducing the consumption of resources and the cost of sterilization.

Introduction Shah and Bhargava 2017). Thus, “as of July 15, 2019, the
FDA sponsors two public innovation challenges on their
The use of ozone in water treatment is common since the
website to identify alternatives to ethylene oxide steriliza-
beginning of the 20th century (Gomella 1972). From there,
tion methods”. Although vapor-phase hydrogen peroxide
its utilization has been enlarged to air sanitization and food
(VHP) sterilizers seem to be the new trend in low-
industry applications (Kim, Yousef, and Dave 1999; Kim,
temperature sterilization, they have their own disadvan-
Yousef, and Khadre 2003). A recently developed field is
tages and limitations such as their expensiveness (Rutala
ozone-based disinfection and sterilization of medical
and Weber 2008; Shah and Bhargava 2017). Additionally,
devices which became conceivable and manageable only
over the last years, global health-care expenditure has
since the beginning of the 21st century. Already in 1987,
increased and public- as well as private-health systems
Rickloff described the benefits of ozone for sterilization of
face rising costs, forcing them to rethink their politics and
thermo-sensitive medical equipment thanks to its broad
become more cost-efficient (Allen, 2019; OECD 2015).
antimicrobial properties and sporicidal activity at low tem-
A trend toward greener technologies further adds pressure
peratures. However, the intrinsic instability of the gas and
to find alternatives for conventional sterilization methods.
the complexity of the kill kinetics made it difficult to
In this context, we have developed and patented an
implement ozone in the highly regulated environment of
environmentally friendly low-cost sterilization technology
hospital sterilization. In 2003, the Canadian company
based on ozone as a sterilizing agent. The strong oxidative
TSO3 Inc. was the first to obtain FDA (US Food and
capability of ozone in both the gaseous form and dissolved
Drug Administration) approval for a sterilizer based on
in water make it ideal for disinfection and sterilization.
a humidified gaseous ozone sterilization technology.
Medical device regulations are extremely demand-
Several developments have led to a growing interest in
ing, and new sterilization technologies must prove their
such alternative sterilization solutions. Including
reliability and effectiveness against the most resistant
a continuous increase in the number of heat-sensitive med-
microorganism. In the case of ozone sterilization,
ical devices which are heavily damaged by conventional
spores of Geobacillus stearothermophilus have been
moist-heat sterilizers either because of their electronic com-
recognized to be the most resistant (de Souza Botelho-
ponents or due to the presence of thermo-sensitive materi-
almeida et al. 2018; Dufresne, Hewitt, and Robitaille
als (e.g. minimally invasive surgical instruments). Existing
2004; Dufresne, Leblond, and Chaunet 2008; Mahfoudh
technologies using ethylene oxide (EtO) have long been
et al. 2010; Ohkawa et al. 2004; Sakurai et al. 2003). The
used as an alternative to steam, but the environmental
selection of the biological indicators and the tests of
and health risks are substantial (Rutala and Weber 2008;
this study were performed following the standards of

CONTACT Sandy A. Thill sandy.thill@sterilux.ch SteriLux SA, Chemin du Viaduc 12, Prilly 1008, Switzerland
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/bose.
© 2019 International Ozone Association
2 S. A. THILL AND M. SPALTENSTEIN

the International Organization for Standardization these parameters affect the inactivation kinetics of
(ISO) on the guidance for the selection, use and inter- ozone (Aydogan and Gurol 2006; Sakurai et al.
pretation of results of biological indicators for the ster- 2003).
ilization of health-care products (ISO 11138–7:2019).
As absolute sterility does not exist, a retreated surgical
instrument is defined as ‘sterile’ when a sterility assur-
Materials and methods
ance level (SAL) of 10−6 is reached (CEN 2001). In
other terms, in 1 million sterilized devices less than The developed sterilization system uses mercury-free exci-
one should present a living germ. However, it is not mer V-UV lamps (172 nm) to generate ozone. Although
possible to directly prove such low probabilities; there- corona discharge is the preferred method for many appli-
fore, extrapolations of survivor curves can be per- cations, it has to be fed with clean and dry air or pure
formed when the characteristics of a sterilization oxygen otherwise formation of nitrogen oxides and corro-
process fulfill certain criteria. sion can occur (Kogelschatz 2003). This is not the case for
The purpose of the developed system is to sterilize V-UV ozone generation (Eliasson and Kogelschatz 1991),
efficiently at ambient temperature without a chamber that and additionally, it has the advantage to be efficient in high
controls the temperature and the pressure. Having no humidity environments essential for successful sterilization
pressure change (e.g. no vacuum steps) during a cycle (Ishizaki, Shinriki, and Matsuyama 1986; Mahfoudh et al.
circumvents decalibration of sensors and reduces mechan- 2010; Sakurai et al. 2003). Being able to generate ozone
ical stress on medical instruments. The process solely relies directly in the ambient air of a hermetically closed con-
on the diffusion of ozone throughout the sterilization con- tainer, relying solely on ozone diffusion, facilitates not only
tainer and requires only small quantities of distilled water the design of the sterilizer but also reduces the complexity
(5 mL) to reach saturation of humidity in the container. of the device and improves the operator safety. There is no
This makes the technology not only an environmentally handling of pure oxygen and no need to withstand high-
friendly alternative to steam, EtO or hydrogen peroxide pressure changes linked to vacuum steps as performed by
processes but also a suitable solution for regions of the some authors (de Souza Botelho-almeida et al. 2018;
world suffering from lack of clean water and recurrent Dufresne, Hewitt, and Robitaille 2004; Dufresne, Leblond,
power outages. and Chaunet 2008; Rediguieri et al. 2016).
Furthermore, in the literature, many studies on ozone Moreover, reducing the amount of consumables and
disinfection and sterilization have focused on high gas resources necessary for a sterilization cycle makes these
concentrations of 4,000 to 40,000 ppm (parts per million) technologies advantageous compared to other existing
(de Souza Botelho-almeida et al. 2018; Mahfoudh et al. low-temperature sterilization technologies (e.g. VHP
2010; Masuda et al. 1990; Robitaille et al. 2006; Sakurai and EtO sterilizers). At the moment, our technology
et al. 2003), compared to the relatively low concentrations results in high D-values compared to those found in the
of 100 to 1,000 ppm generated by our process. Moreover, literature. Nevertheless, the above-mentioned advan-
previous work has shown high variability in the inactiva- tages and the possibility to have a sterile storage in
tion rates and D-values (decimal reduction times or doses the hermetically closed container show that this tech-
required, at given conditions (e.g. temperature, ozone con- nology has a huge potential.
centration), to achieve one log reduction, i.e. to kill 90% of
relevant microorganisms) depending on the conditions
used for the studies (Aydogan and Gurol 2006; Sousa
Ozone generation
et al. 2011). Besides, the exact inactivation mechanism of
ozone and the additional oxidative agents that form in the A lamp unit consisting in an airtight metal case with
presence of water vapor are not yet completely understood a quartz window (SK-1300, synthetic fused silica, Ohara
(Cortezzo et al. 2004; Khadre and Yousef 2001; Mahfoudh Corporation) is placed above the quartz (SK-1300, syn-
et al. 2010). The complexity of the mechanisms at play thetic fused silica, Ohara Corporation) of a hermetically
makes it necessary to characterize each new ozone steriliza- closed container made of Polybutylene terephthalate
tion technology and process. Small variations in the con- (PBT). Three 172 nm ozone generation lamps (mercury-
ditions in use can have a huge impact on the outcome. free V-UV lamps, UXFL130-172-Z1, Ushio) are present
Here we would like to prove that the developed inside the lamp unit. The V-UV light from the lamp unit
ozone technology fulfills the criteria given by the passes through the two quartz windows to generate ozone
standards for medical device sterilization. The ster- from the ambient air inside the container. The tests were all
ilization efficiency was characterized in the intended performed inside the same container and with the same
temperature and humidity range as it is known that lamp unit (Figure 1).
 OZONE: SCIENCE & ENGINEERING 3

Figure 1. (a) The container, which was used for the tests, is made of PBT. A quartz window in the lid allows the UV light to enter it.
(b) The lamp unit consisting in a metal case and a quartz window was used to generate ozone and to measure the ozone
concentration. Three-generation UV lamps, as well as the measurement system (LED and sensor), were present inside this unit.

Ozone measurement system and calibration Measurements were performed every 30 s to have
a precise control of the concentration.
The custom-built optical measurement system consists in
a LED (MTE280H33-UV, Marktech Optoelectronics)
that emits 280 nm UV light through the quartz inside
Test setup
the container. The ray is reflected by a polyvinylidene
difluoride (PVDF) surface and then measured by a SiC- Process challenge devices (PCDs) consisting in metal
based UV sensor with an integrated amplifier and voltage tubes with a length of 10 cm and a diameter of 3 mm
output (TOCON_ABC3, sglux GmbH). The calibration were used (Figure 2(a)). For each test, four BIs were put
of the measurement system was performed using an in the middle of these PCDs which were then placed in
external ozone sensor (Ozone Analyzer BMT 964, BMT the most “difficult-to-sterilize” site of the container (i.e.
Messtechnik GmbH). on top of a basket as seen in Figure 2(b) – side view and
Figure 2(c) – top view). For each test, a stainless-steel
basket filled with medical stainless-steel tools (total
weight 7.5 kg) was placed inside the container.
Software
5 mL of distilled water was added on a 300g/m2 blotting
An Arduino Uno board was used to control the measure- paper (dimensions 160 mm x 60 mm, Zuber Rieder) before
ment system and the ozone generation by turning the UV the container was closed and a test was launched.
lamps on and off. To obtain stable ozone concentration, the The tests were performed inside a custom-built cli-
lamps were turned on for specific time intervals as soon as mate chamber allowing tests to be performed at fixed
the measured concentration fell below fixed thresholds. temperatures (15 °C, 20 °C, 23 °C, 25 °C and 30 °C).

Figure 2. (a) The BIs were placed in PCDs consisting in metal tubes. (b) Side view and (c) top view of the container and the
placement of the PCDs used for testing the SterOx process. (d) Zoom of the placement of the PCDs.
4 S. A. THILL AND M. SPALTENSTEIN

Bacterial strain The majority of commercially available ozone sen-

sors and analyzers use optical sensors to measure the
Commercially available paper carrier BIs inoculated with
ozone concentration, which is often given in ppm.
Geobacillus stearothermophilus spores (ATCC #7953) were
Sensors accurate over a wide range of concentrations
obtained from Mesa Laboratories Inc. (3 batches) and from
are rare and mainly require an intake of the medium to
CrossTex (1 batch). The dimensions of the BIs were the
be analyzed. It was therefore decided to develop
same: 2 mm x 10 mm. The first three batches were bought
a noninvasive optical system based on the absorption
from Mesa Laboratories: BI 1 (lot # CGS-309) had
characteristics of ozone.
a nominal population of 2.5 × 106 spores/carrier, BI 2 (lot
Ozone absorbs UV light with a peak at 254 nm until
# CGST-330) had a nominal population of 2.6 × 106
around 300 nm (Malicet et al. 1995). The more ozone is
spores/carrier, and BI 3 (lot # CGST-331) had a nominal
present in the container, the stronger is the absorption of
population of 1.9 × 106 spores/carrier. The last batch
the 280 nm light ray emitted by an LED leading to a lower
S1534301 of 2.6 × 106 spores/carrier was bought from
signal received by the sensor. Additionally, the absorption
CrossTex and was referenced as BI 4. The initial spore
spectrum of hydroxyl radicals (230–320 nm) seems to
population before sterilization of each BI batch was con-
indicate that it is possible that at 280 nm not only the
trolled by using an unprocessed BI as a positive control.
concentration of ozone is measured, but also partly the
The results of the positive controls were between 49-53% of
concentration of HO˙ molecules, which play an important
the nominal populations (N0) given by the manufacturer,
role in the sterilization process (Chipman 2008; Mahfoudh
indicating that the method for the recovery of the spores
et al. 2010).
was suitable since the acceptance criterion given by ISO
The calculation of the ozone concentration is based
11138:7–2019 for the recovered population of a positive
on the Beer–Lambert law (Swinehart 1962; Wilson and
control is 50-300% of N0 (i.e. the recovered population can
Birks 2006) which relates the attenuation of a light ray
vary significantly from one operator to another and is also
to the properties of the analyte (in this case ozone)
depending on the protocol and the plating techniques
present in the path of the light.
used).
The accuracy and repeatability of this sensor being
crucial for the whole technology, it had to be calibrated
Enumerations with a commercially available external ozone analyzer. The
repeatability of the system was more important than its
The enumerations were performed on Trypticase soy agar
accuracy, because having different concentrations instead
plates (90 mm, Ref. 43019, bioMérieux). After each experi-
of the targeted 550 ppm would not have been a problem as
ment, the BIs were transferred into 15 mL tubes containing
long as the value remained constant during all the test
3 mL of sterile distilled water and 5 glass beads with
series (i.e. no variations from one experiment to the
a diameter of 5 mm. The tubes were vortexed for 5 to 8
other). Nevertheless, the accuracy was key to be able to
min until the BIs were macerated to pulp. Serial dilutions
compare the results of the microbial tests with those found
were performed before spreading 150 to 600 µL per agar
in the literature.
plate to obtain 30 to 300 colony forming units (CFUs). The
The theoretical calculations were verified and cor-
plates were incubated at 57 °C for 20 hours (±2 hours)
rected by calibrating the system with the external ozone
before counting the CFUs. Waiting 48 hours before count-
sensor to ensure its accuracy in the targeted range of
ing resulted in an increase of less than 10% of the total
100 to 1,000 ppm.
number of CFUs per plate, leading to a neglectable increase
The ratio of the signal measured with the ozone analyzer
in the survivor population. The analysis of the results was
and the developed measurement system, shown in Figure 3,
performed according to the survival curve method
was constantly 1.1 over the targeted concentration range
described in ISO 14161:2009 (replaced by ISO 11138–
with a coefficient of determination, r2, of 0.9974. Thus,
7:2019), allowing the calculation of the D-value of the
a correction coefficient of 1.1 was included (i.e. the mea-
survivor curves.
sures of the system had to be multiplied with 1.1) to obtain
measurements as close as possible to the actual ozone con-
Results centration. Additionally, a temperature correction was
applied due to the temperature dependency of the LED’s
Ozone measurement system
emission (results not shown). The increase of the lamp
In order to show the efficiency of this ozone-based unit’s temperature during the ozone generation phases as
sterilization process, it is crucial to measure the ozone well as temperature changes during a cycle make it impor-
concentration. tant to add such a correction.
 OZONE: SCIENCE & ENGINEERING 5

Figure 3. Results of the calibration test of the developed ozone measurement system with a BMT 964 ozone analyzer, r2 = 0.9974.

After having ensured that the measurement system sterilization processes (Mahfoudh et al. 2010; Masuda et al.
was accurate, the tests at constant ozone concentrations 1990; Sakurai et al. 2003), which states that the sterilization
were started. Intermediate stability tests were per- is not only due to ozone in itself but also to oxidative agents
formed on a regular basis to guarantee the repeatability that form when ozone molecules react with water mole-
of the measurement system, which was estimated to be cules. In order to guarantee a high RH in the container for
±30 ppm (results not shown). each test, 5 mL of distilled water was added on a blotting
paper to facilitate evaporation. The humidity increase to
>95% lasted 1–2 hours, depending on the temperature and
Parameters affecting ozone sterilization
the ambient RH.
Absolute sterility cannot be achieved; hence, goods may Furthermore, the PCDs with the biological indica-
be labeled terminal sterile, according to the European tors were always placed in the most “difficult-to-
Standard EN 556, if a minimum SAL of at least 10−6 sterilize” site of the container. As ozone is heavier
CFU/per part is reached. The overkill method consists than air, there is a small concentration gradient inside
in starting at an initial population of 106 spores/per BI the container (i.e. the ozone concentration in the upper
which is higher than the expected bioburden. Thus, to part is lower than at the bottom of the container).
reach a SAL of 10−6, a 12 log reduction has to be proven. Moreover, the water was added in the middle of the
At first, enumeration tests have to be performed to show box, leading to a minimal RH in the corners of the
0.5–4 log reductions (LR) (also known as the survivor container. Tests had been performed to verify these
curve method). In a second step, the results can be extra- theoretical assumptions (results not shown), and the
polated to 12 LR, given that the coefficients of determina- most “difficult-to-sterilize” site was found to be as
tion of the survivor curves are >0.8 (criterion defined in expected in the upper corners of the container. For
ISO 11138:7–2019). each test, four BIs were used and the results were
Several series of tests at different conditions were averaged to obtain one datapoint per test (as described
performed to define the survivor curves of our process. in ISO 11138–7:2019).
The main purpose of these tests was to validate the For each test series performed at a fixed tempera-
sterilization process and to prove its efficiency at ambi- ture, the only variable was the ozone dose, correspond-
ent temperatures (i.e. 15 °C–30 °C). ing to the ozone concentration in contact with the
Preliminary experiments had been performed to narrow sterilization load during the exposure time. Few studies
the conditions for a successful sterilization, including some mention the ozone dose and most only express the
humidity studies showing that a high relative humidity survivor curves in the function of exposure time,
(RH) is required for the sterilization to be successful. This increasing the difficulty to compare D-values as varying
is in agreement with the current understanding of ozone ozone concentrations are being used. We decided to
6 S. A. THILL AND M. SPALTENSTEIN

plot the survivor curves in the function of the ozone Mahfoudh et al. 2010), whereas here, single-phase survivor
dose, which we unconventionally expressed in ppm curves were obtained. Heterogeneities in spore populations
times hours (ppm*h). or clumping of spores protecting a small percentage of
them could have caused these phenomena in the survivor
curves of earlier studies. Interestingly, a lag phase is present
Influence of temperature
before the exponential decrease of the spore populations.
The temperature’s influence on the sterilization process
Aydogan and Gurol (2006) as well as Ishizaki, Shinriki, and
had to be studied because our system does not control it.
Matsuyama (1986) have recorded similar lag phases for
The setup was therefore placed in a climate chamber to
bacillus spores before the start of the actual inactivation
perform tests at different constant temperatures. The aim
phase. It has been proposed that a certain minimal damage
was to remain around possible ambient temperatures (15 °
is required in the spore coat before inactivation can occur
C, 20 °C and 30 °C). During these tests, the ozone concen-
(Aydogan and Gurol 2006; Hoffman and Spiner 1970;
tration was constantly around 550 ppm (±7 ppm). The
Yokoya and York 1965). We defined the lag phase as the
constant amount of ozone was reached by regenerating
ozone dose necessary for reaching 0.5 LR. This corresponds
ozone each time the ozone concentration fell below pre-
to a threshold from which we clearly observed exponential
defined thresholds, allowing oscillations around the tar-
decreases of the spore populations.
geted value.
Validation of the linearity of a sterilization process
requires the coefficient(s) of determination, r2, to be Influence of BI batch
above 0.8 (cf. ISO 11138–7:2019). The r2 of each survivor Additional tests, with different BI batches, were performed
curve in Figure 4 is >0.9 meaning that the inactivation at intermediate temperatures: 23 °C and 25 °C. After these
reactions at each tested temperature follow first-order log- tests, a fourth batch of BIs from a different manufacturer
linear kinetics. Thus, an extrapolation of the results to 12 was used to repeat the tests at 30 °C, allowing a direct
LR can be applied. Note that intermediate 6–8 log reduc- comparison of the resistance characteristics of different BI
tions can be obtained with the fraction negative method batches (Figure 5). The survivor curves followed the same
and that such tests were performed. The results were trend as those in the previous tests. However, some char-
coherent with those of the survivor curve method and acteristics seem to depend more on the BI batch than on
were therefore not added to this study. external conditions such as temperature and humidity.
In the literature, two-phase survival curves have been The decimal reduction factor, also known as D-value,
observed by different authors (Aydogan and Gurol 2006; represents the resistance of a specific germ (e.g. BI batch)
Cerf 1977; Kowalski, Bahnfleth, and Whittam 1998; for a sterilization process under defined conditions. This

Figure 4. Influence of temperature on the survivor curves. The r2 were 0.975, 0.981 and 0.901 for the 15 °C, 20 °C and 30 °C curves,
respectively. Each datapoint and its error bars represent the mean of four replicates used per test.
 OZONE: SCIENCE & ENGINEERING 7

Figure 5. Influence of BI batch on the survivor curves.

D-value can be obtained from the slopes of the survivor its lag phase, whereas the D-value fits in the previous
curves and corresponds to the time or dose necessary to values.
obtain 1 log10 reduction. The D-values, obtained for the
data in Figure 4 as well as for the additional temperatures, Influence of ozone dose
are plotted in Figure 6. Verifications were performed to ensure that the use of the
A total of six test series had been performed and the ozone dose concept is suitable for characterizing an ozone-
D-values, as well as the ozone doses necessary to reach based sterilization process. Among others, the results of
0.5 LR, are given in Table 1. As already mentioned, Mahfoudh et al. (2010) suggest that the ozone dose could
batch 4 was bought from a different manufacturer than be an alternative (or an even better characterization criter-
batches 1–3 and important differences can be seen for ion) to the usually used exposure time. Moreover, ISO

Figure 6. Relationship between D value and temperature, with r2 = 0.9110 for BI batches of the same manufacturer.
8 S. A. THILL AND M. SPALTENSTEIN

Table 1. Ozone doses necessary to reach 0.5 LR (lag phase) and and therefore confirm that it is possible to use our
D-values of the survivor curves at different temperatures using system for sterilizing medical devices at ambient tem-
different BI batches. peratures. This finding is consistent with data from
Temp. [°C] Lag phase [ppm*h] D-value [ppm*h] BI batch #
studies using a similar temperature range (Aydogan
15 3315 570 2
20 2435 480 2 and Gurol 2006; Sakurai et al. 2003).
23 2155 495 3 Moreover, the overkill method is appropriate for
25 1370 470 1
30 2180 400 2 validating this technology as the survivor curves follow
30 1560 555 4 first-order log-linear kinetics.

11138–7:2019 explicitly mentions that for some steriliza-

tion processes based on radiation or ozone, the resistance D-value
characteristics can be expressed in dose instead of time.
The r2 of 0.9110 observed for the data in Figure 6 allows to
The tests of this study were performed at constant ozone
conclude that in this range of temperatures, the D-value
concentrations. Using the ozone dose instead of the expo-
follows a linear regression. This is in line with literature on
sure time has, therefore, no direct impact on these char-
the influence of temperature on ozone sterilization
acterization results, since the ozone dose is defined as the
(Sakurai et al. 2003). It was observed that the D-value
integration of the ozone concentration by the exposure
decreases (i.e. increase of the inactivation rate) with
time. Experimentally, this is translated in the concentration
increasing temperature. This phenomenon matches with
at each measurement point multiplied by the time interval.
the Arrhenius equation which states that the rate of such
In order to study the concept of ozone dose, two test
chemical reactions is higher when the temperature
series consisting in experiments at 138 ppm, 275 ppm
increases, explaining the faster sterilization at 30 °C than
and 550 ppm were performed at 23 °C and 30 °C,
at 15 °C. However, higher temperatures also lead to faster
respectively (Figure 7). The above-mentioned lag
spontaneous decomposition of ozone to oxygen (natural
phases and D-values are the important characterization
exponential degradation processes linked to the instability
criterions that are compared in Table 2.
and high reactivity of ozone as well as to the presence of
high RH levels). Reactions between different species pre-
Discussion sent in the container (e.g. O3, O2, O(1D), OH˙ and H2
O molecules) lead to the reduction of the ozone density
Feasibility to sterilize at ambient
(Ono et al. 2014). At 15 °C, the ozone half-life for an empty
The results of the tests performed at 15 °C-30 °C container, in which 5 mL of distilled water was added, was
allowed to apply an extrapolation to 12 log reductions around 16 hours; at 30 °C, it was approximately 6 hours;

Figure 7. Influence of the ozone concentration on the survivor curves at 23 °C and at 30 °C.
 OZONE: SCIENCE & ENGINEERING 9

Table 2. Ozone doses necessary to reach 0.5 LR (lag phases) internal biomolecules. This is in line with findings of
and D-values for the survivor curves of three ozone concentra- Cortezzo et al. (2004) and of Young and Setlow (2004)
tions at two different temperatures. stating that oxidizing agents such as ozone weaken and
30 °C 23 °C
inactivate spores by damaging their inner membrane. The
Ozone D-value Lag phase D-value Lag phase
[ppm] [ppm*h] [ppm*h] [ppm*h] [ppm*h] spore coat that normally protects the spores from harm
550 555 1560 495 2155 could be weakened by swelling, thus facilitating the entry of
275 540 820 405 1765 radicals and ozone, which can then attack the inner mem-
138 485 724 450 1190
brane. Khadre and Yousef (2001) also concluded that
ozone might attack the outer spore coat layer. Moreover,
and at 40 °C, it was less than 4 hours. Therefore, more Aydogan and Gurol (2006) hypothesized about the pre-
ozone has to be produced at higher temperatures to equili- sence of micro-condensations through the spore’s surface
brate the ozone decomposition, which might not be realiz- which would allow ozone dissolution and would increase
able with the current developed sterilizer. Temperatures its biocidal action.
above 35 °C were tested (results not shown), but 550 ppm For our setup, adding 5 mL of water at the beginning of
could not be generated in a repeatable manner, yet (a setup each test ensures an increase to >95% RH. Preliminary tests
with a better ozone generating performance is currently with humidity sensors inside a container showed that the
under development to further expand the validated tem- increase in RH lasts around 1 to 2 hours (depending on the
perature range). However, it seems that lower temperatures temperature). This corresponds to an ozone dose of 550 –
are sufficient to achieve the goal to sterilize. 1,100 ppm*h since the tests were performed at constant
ozone concentrations of 550 ppm. However, the dose to
reach 0.5 LR was considerably higher as seen in Table 1
Lag phase
(1,300 – 3,300 ppm*h). It is possible that the humidity
It was observed that the bacterial population remained diffusion into the PCDs where the BIs were located took
constant for some time, clearly showing a latency between longer than what has been measured, explaining the longer
the starting of the sterilization process and the death of lag phases.
bacteria. This interval was defined to be the ozone dose Unlike the D-values, the lag phases in function of
necessary to reach 0.5 log reduction. Aydogan and Gurol temperature did not follow a linear regression (Figure
(2006) demonstrated that pre-hydration of the spores dur- 8). Even for the same batch of BIs, a high variation
ing 3 hours at high RH can nearly completely eliminate the could be observed in function of the temperature. This
lag phase. It was therefore supposed that this lag phase indicates that other parameters influence the time or dose
could be due to the relatively slow increase of humidity at it takes before the bacterial population starts decaying.
the beginning of each test, whereas other authors start at These parameters can be intrinsic such as differences
a high initial humidity or with a much faster increase of RH between BI batches or extrinsic concerning the experi-
(de Souza Botelho-almeida et al. 2018; Dufresne, Hewitt, mental setup conditions.
and Robitaille 2004; Dufresne, Leblond, and Chaunet Moreover, the very short lag phase of the curve at 25 °C
2008). led to a significant shift to the left of the curve, which was
Ozone sterilization processes along with the resulting different from the expected trend seen for the previous
inactivation of microorganisms are not solely based on the tests, but the D-value fitted in between the other values.
oxidizing power of ozone, which favors organic com- Due to the natural variability of some uncontrolla-
pounds, but also on the nonselective oxidizing power of ble parameters, such as relative humidity in the
the highly reactive hydroxyl and hydroperoxyl radicals laboratory as well as the ambient temperature, the
(Kim, Yousef, and Dave 1999; Mahfoudh et al. 2010; time to reach the targeted temperature in the climate
Sakurai et al. 2003). In the presence of humidity, two chamber varied from 5 to 30 min. During this time,
main reactions lead to the formation of OH˙ radicals: (1) the ambient humidity in the container could have
V-UV light splits H2O molecules and (2) O(1D) reacts with varied before the water was added and the test was
H2O; allowing further reactions leading to the formation of launched (i.e. ozone generation started). Nevertheless,
HO2˙ radicals (Ono et al. 2014; Salvermoser, Kogelschatz, the high coefficients of determination show that the
and Murnick 2009). variations in the environmental conditions were prob-
The tests of Mahfoudh et al. (2010) with dry (<2% RH) ably neglectable, else less coherent results would have
and humidified ozone (>70% RH) give new insights in the been obtained.
sterilization mechanism of ozone: They noted that swelling It was possible to rule out some intrinsic factors because
of spores under humidified media might lead to a better BIs of the same size were used, which had similar D121-
penetration of biocidal agents improving reactions with values for steam sterilization (i.e. the time it takes to reach 1
10 S. A. THILL AND M. SPALTENSTEIN

Figure 8. Lag phases obtained for different BI batches in function of the temperature.

log reduction for steam cycles performed at 121 °C). As always have to indicate the D-value for specific
there are nearly no ozone sterilization processes, BIs man- conditions.
ufactured specifically for this application are not available However, the observed important differences make it
and those for steam sterilization controls were bought. problematic to correctly define a sterilization cycle that
always reaches at least 12 log reductions independent of
the BI used. Similar issues have been recognized by Reich
Influence of BI batch
and Caputo (2004) for VHP BIs. We decided to base our
For VHP sterilization processes, the same species of sterilization cycle upon the longest lag phase that was
G. stearothermophilus spores can result in BIs with varying observed. Additionally, a temperature dependency was
resistance characteristics depending on, for example, the included to adjust the ozone dose necessary for
method of evaluation, the packaging or the carrier sub- a sterilization cycle, based on the ambient temperature.
strate (Eric 2014; Reich and Caputo 2004; Yokoya and Furthermore, several authors have encountered pro-
York 1965). This might also be true for the resistance to blems comparing their results to previous studies due
ozone sterilization. Ozone sterilization relies on similar to various test conditions (de Souza Botelho-almeida
oxidation processes as VHP and is also a surface steriliza- et al. 2018; Rickloff 1987; Sousa et al. 2011). Our results
tion contrary to hot steam methods. show that several parameters affect the sterilization
The D-value seems to be less affected by internal and process of ozone considerably and that already the
external parameters than the lag phase. The BIs from choice of the biological indicator is crucial and can be
the same manufacturer (BIs 1–3) show, as expected, the cause of some of the observed variations.
a decrease of the D-value with increasing temperature. This problem to reproduce and compare results
However, the D-value for batch 4 was aberrant (Figure represents one of the major challenges of ozone ster-
6) which could be due to intrinsic differences in the ilization technologies and makes it difficult to put in
way the manufacturers produce the BIs. Variations of place regulations that apply to all ozone sterilizers.
the methods for sporulation and inoculation as well as Nevertheless, the high potential not only for medical
for production of the carrier materials can lead to device sterilization but also for sterilization in more
significant differences in BI resistance (Eric 2014; recent fields such as tissue engineering and newly
Reich and Caputo 2004). developed implantable scaffolds and hydrogels
This could explain the observed discrepancies in (Galante et al. 2017; Rediguieri et al. 2016), make
D-values of BI batches from different manufacturers. ozone gas a highly promising sterilization agent. More
The fact that each BI batch has an individual resistance efforts are still required to improve the understanding
to a sterilization process is not new (Stanbury, of the sterilization mechanisms at work and the para-
Whitaker, and Hall 2013) and manufacturers of BIs meters that influence them.
 OZONE: SCIENCE & ENGINEERING 11

Ozone dose ozone compatibility and resistance of many materials less

problematic. The next step will be to test other materials
For the dose verification tests, we expected to see that
and more complex geometries of surgical tools (e.g. hollow
the ozone dose required to reach a certain log reduction
tubes). Ozone compatibility will play a crucial role in these
would be identical for the three concentrations.
future studies. It is known that some materials are strongly
However, for the tests at 23 °C and for those at 30 °C,
degraded by ozone as, for example, nitrile and natural
the required dose decreased with decreasing ozone
rubber, whereas others can be sterilized as easily as stainless
concentration. The exposure time seems to be more
steel (e.g. PTFE, silicon, glass). The ozone concentration to
important than the ozone concentration; using a twice
which the materials are exposed plays an important role in
as high concentration did not decrease the sterilization
defining their compatibility. And it is clearly advantageous
time by a factor two. Aydogan and Gurol (2006) per-
to have lower ozone concentrations and not as high ones as
formed tests at 500 ppm and at 1000 ppm and observed
used in other studies.
a decrease of the D-value from 79 min to 65 min.
Additionally, Mahfoudh et al. (2010) demonstrated
a decrease from 12 min to 8.3 min as they doubled Conclusion
their ozone concentration from 2000 ppm to 4000 ppm.
This indicates that over a wide range of concentrations, In this study, we have demonstrated that this is possible to
the D-value is not directly depending on the ozone sterilize medical device using a low concentration of ozone
concentration. Although it seems possible to decrease as well as ambient temperatures and small quantities of
the D-value by increasing the ozone concentration. water. This is an important step toward a better under-
Aydogan and Gurol (2006) observed a threshold of standing of ozone sterilization technologies, which is in line
1500 ppm above which an increase of the concentration with the new challenges that sterilization faces due to the
has no effect on the D-value and they explained it with continuous increase in the complexity of surgical instru-
a possible limitation by the rate of mass transfer of ments and devices. A technology based on ozone as
ozone. However, it is possible that their observation is a sterilization agent has the potential to overcome major
partly due to a less concentration-dependent minimal problematics and become an affordable as well as an eco-
damage to the spores before inactivation starts (as friendly alternative to EtO or hydrogen peroxide processes.
mentioned earlier for the lag phase). No toxic residues or by-products remain after ozone ster-
Despite these findings, we decided to continue ilization (i.e. complete decomposition of ozone to oxygen)
expressing the D-value in ozone dose instead of expo- and only low amounts of resources are required to perform
sure time. The exposure time seems to be as inadequate a sterilization cycle.
as the dose because it does not allow to compare the
results of several studies performed at different ozone Acknowledgments
concentrations. By validating our process at the highest
ozone concentration, it can be guaranteed that when The authors would like to thank the whole team at SteriLux
SA, especially Lucas Meyer and Chloé Pacquier, for their
lower concentrations occur during a sterilization cycle,
support. Thanks to Nathalie Meyer for her editorial support.
the dose will be sufficient to reach a SAL of at least
10−6. Ozone concentrations above 550 ppm would
require additional validation tests to those performed Conflicts of Interest
for this study. Further studies are required for better M.S. has interests in the intellectual property described
understanding the effect of ozone concentration on the herein. S.A.T. and M.S. have financial interests in SteriLux
inactivation rate of micro-organisms. SA.

Future steps ORCID

Sandy A. Thill http://orcid.org/0000-0003-0012-7768
Our results clearly show that some parameters consider- Marc Spaltenstein http://orcid.org/0000-0003-3725-0276
ably influence the outcome of sterilization processes using
low ozone concentrations. Nevertheless, the understanding
of the process is sufficient to prove the efficiency of ozone References
to sterilize medical devices. Besides, the feasibility to ster-
Allen, Stephanie. 2019. “2019 Global Health Care Sector
ilize successfully in a temperature range of 15 °C to 30 °C Outlook.” Deloitte https://www2.deloitte.com/global/en/
opens the door for sterilization of heat-sensitive medical pages/life-sciences-and-healthcare/articles/global-health-
devices. The low ozone concentrations used here make the care-sector-outlook.htmL
12 S. A. THILL AND M. SPALTENSTEIN

Aydogan, A., and M. D. Gurol. 2006. “Application of Gaseous Kim, J.G., A. E. Yousef, and S. Dave. 1999. “Application of
Ozone for Inactivation of Bacillus Subtilis Spores.” Journal Ozone for Enhancing the Microbiological Safety and
of the Air & Waste Management Association 56 (2):179–85. Quality of Foods: A Review.” Journal of Food
doi:10.1080/10473289.2006.10464443. Protection 62 (9):1071–87. doi:10.4315/0362-028X-
CEN. 2001. “EN 556-1 Sterilization of Medical Devices - 62.9.1071.
Requirements for Medical Devices to Be Designated Kim, J.G., A.E. Yousef, and M.A. Khadre. 2003. “Ozone
“Sterile” - Part 1: Requirements for Terminally Sterilized and Its Current and Future Application in the Food
Medical Devices.” October, 1. Industry.” Advances in Food and Nutrition Research
Cerf, O. 1977. “A Review Tailing of Survival Curves of 45:167–218.
Bacterial Spores.” Journal of Applied Bacteriology 42 Kogelschatz, Ulrich. 2003. “Dielectric-barrier Discharges:
(1):1–19. doi:10.1111/jam.1977.42.issue-1. Their History, Discharge Physics, and Industrial
Chipman, Daniel M. 2008. “Absorption Spectrum of OH Applications.” Plasma Chemistry and Plasma Processing
Radical in Water.” The Journal of Physical Chemistry A 23 (1):1–46. doi:10.1023/A:1022470901385.
112 (51):13372–81. doi:10.1021/jp807399b. Kowalski, W. J., W. P. Bahnfleth, and T. S. Whittam. 1998.
Cortezzo, D. E., K. Koziol-Dube, B. Setlow, and P. Setlow. Bactericidal effects of high airborne ozone concentra-
2004. “Treatment with Oxidizing Agents Damages the tions on Escherichia coli and Staphylococcus aureus.
Inner Membrane of Spores of Bacillus Subtilis and 205–21.
Sensitizes Spores to Subsequent Stress.” Journal of Mahfoudh, A., M. Moisan, J. Seguin, J. Barbeau, Y. Kabouzi,
Applied Microbiology 97 (4):838–52. doi:10.1111/j.1365- and D. Keroack. 2010. “Inactivation of Vegetative and
2672.2004.02370.x. Sporulated Bacteria by Dry Gaseous Ozone.” Ozone:
de Souza Botelho-almeida, T., F.R. Lourenço, I.S. Kikuchi, Science & Engineering 32 (3):180–98. doi:10.1080/
R. Awasthi, K. Dua, and T.D.J.A. Pinto. 2018. “Evaluating 01919511003791971.
the Potential, Applicability, and Effectiveness of Ozone Malicet, J., D. Daumont, J. Charbonnier, C. Parisse,
Sterilization Process for Medical Devices.” Journal of A. Chakir, and J. Brion. 1995. “Ozone UV Spectroscopy.
Pharmaceutical Innovation 13 (2):87–94. doi:10.1007/ II. Absorption Cross-sections and Temperature
s12247-017-9308-7. Dependence.” Journal of Atmospheric Chemistry 21
Dufresne, S., A. Hewitt, and S. Robitaille. 2004. “Ozone (3):263–73. doi:10.1007/BF00696758.
Sterilization: Another Option for Healthcare in the 21st Masuda, S., E. Kiss, K. Ishida, and H. Asai. 1990. “Ceramic-
Century.” American Journal of Infection Control 32 (3): based Ozonizer for High-speed Sterilization.” IEEE
E26–E27. doi:10.1016/j.ajic.2004.04.038. Transactions on Industry Applications 26 (1):36–41.
Dufresne, S., H. Leblond, and M Chaunet. 2008. doi:10.1109/28.52671.
“Relationship between Lumen Diameter and Length OECD. 2015. Fiscal Sustainability of Health Systems: Bridging
Sterilized in the 125L Ozone Sterilizer.” American Journal Health and Finance Perspectives Paris: OECD Publishing.
of Infection Control 36 (4):291–97. doi:10.1016/j. doi:10.1787/9789264233386-en.
ajic.2007.06.006. Ohkawa, H., T. Akitsu, K. Ohtsuki, M. Ohnishi, and M. Tsuji.
Eliasson, B., and U. Kogelschatz. 1991. Ozone generation 2004. “High-Grade Disinfection Using High-Density
with narrow–band UV radiation. 365–73. Ozone.” Journal of Advanced Oxidation Technologies 7
Eric, Gillitzer. 2014. “Parameters Effecting Vapor Hydrogen (2):154–160. doi:10.1515/jaots.2004.0210.
Peroxide BI Performance.” Spore News 9 (3) July 31 Ono, R., Y. Nakagawa, Y. Tokumitsu, H. Matsumoto, and
https://biologicalindicators.mesalabs.com/wp-content T. Oda. 2014. “Effect of Humidity on the Production of
/uploads/sites/31/2014/07/Spore-News-Vol-9-No3.pdf Ozone and Other Radicals by Low-pressure Mercury
Galante, R., D. Ghisleni, P. Paradiso, V. D. Alves, T. JA Pinto, Lamps.” Journal of Photochemistry and Photobiology A:
R. Colaço, and A. P. Serro. 2017. “Sterilization of Chemistry 274:13–19. doi:10.1016/j.
Silicone-based Hydrogels for Biomedical Application jphotochem.2013.09.012.
Using Ozone Gas: Comparison with Conventional Rediguieri, C.F., T. de Jesus Andreoli Pinto, N.A. Bou-
Techniques.” Materials Science and Engineering: C Chacra, R. Galante, G.L.B. de Araújo, T. Do
78:389–97. doi:10.1016/j.msec.2017.04.073. Nascimento Pedrosa, S.S. Maria-Engler, and P.A. De
Gomella, Cyril. 1972. “Ozone Practices in France.” Journal- Bank. 2016. “Ozone Gas as a Benign Sterilization
American Water Works Association 64 (1):39–45. Treatment for PLGA Nanofiber Scaffolds.” Tissue
doi:10.1002/j.1551-8833.1972.tb02629.x. Engineering Part C: Methods 22 (4):338–347.
Hoffman, R. K., and D. R. Spiner. 1970. “Effect of Relative doi:10.1089/ten.tec.2015.0298.
Humidity on the Penetrability and Sporicidal Activity of Reich, R.R., and R.A. Caputo. 2004. “Vapor-phase Hydrogen
Formaldehyde.” Applied and Environmental Microbiology Peroxide Resistance of Environmental Isolates.”
20 (4):616–19. Pharmaceutical Technology 28 (8):50–59.
Ishizaki, K., N. Shinriki, and H. Matsuyama. 1986. Rickloff, James R. 1987. “An Evaluation of the Sporicidal
“Inactivation of Bacillus Spores by Gaseous Ozone.” Activity of Ozone.” Applied and Environmental
Journal of Applied Bacteriology 60 (1):67–72. doi:10.1111/ Microbiology 53 (4):683–86.
jam.1986.60.issue-1. Robitaille, S., M. Simard, S. Fournier, S. Dufresne, and
Khadre, M. A., and A. E. Yousef. 2001. “Sporicidal Action of R. Turcot. 2006. “Method and Apparatus for Ozone
Ozone and Hydrogen Peroxide: A Comparative Study.” Sterilization.” U.S. Patent 7,128,872, issued October 31.
International Journal of Food Microbiology 71 (2–3):131– Rutala, W. A., and D. J. Weber. 2008. “Guideline for
38. doi:10.1016/S0168-1605(01)00561-X. Disinfection and Sterilization in Healthcare Facilities,
 OZONE: SCIENCE & ENGINEERING 13

2008.” Centers for Disease Control and Prevention (CDC). Literature Review.” Revista Da Escola De Enfermagem Da
https://www.cdc.gov/infectioncontrol/guidelines/disinfec USP 45 (5):1243–49. doi:10.1590/S0080-62342011000
tion/index.htmL 500030.
Sakurai, M., R. Takahashi, S. Fukunaga, S. Shiomi, Stanbury, PF, A Whitaker, and SJ. Hall. 2013. Principles of
K. Kazuma, and H. Shintani. 2003. “Several Factors Fermentation Technology. Kidlington: Elsevier.
Affecting Ozone Gas Sterilization.” Biocontrol Science 8 Swinehart, D. F. 1962. “The Beer-lambert Law.” Journal of
(2):69–76. doi:10.4265/bio.8.69. Chemical Education 39 (7):333. doi:10.1021/ed039p333.
Salvermoser, M. J., U. Kogelschatz, and D. E. Murnick. 2009. Wilson, K. L., and J. W. Birks. 2006. “Mechanism and
“Influence of Humidity on Photochemical Ozone Elimination of a Water Vapor Interference in the
Generation with 172 Nm Xenon Excimer Lamps.” The Measurement of Ozone by UV Absorbance.”
European Physical Journal Applied Physics 47 (2):22812. Environmental Science & Technology 40 (20):6361–67.
doi:10.1051/epjap/2009063. doi:10.1021/es052590c.
Shah, S. B., and A. K. Bhargava. 2017. “Recent Advances in Yokoya, Fumio, and George K. York. 1965. “Effect of Several
Low Temperature sterilization-Moving Ahead from Environmental Conditions on the “Thermal Death Rate”
Cidex™/ETO to OPA/Ozone: An Update.” Indian Journal of Endospores of Aerobic, Thermophilic Bacteria.” Applied
of Anaesthesia 61 (10):855. doi:10.4103/ija.IJA_281_17. and Environmental Microbiology 13 (6):993–99.
Sousa, C. S., L M. Torres, Marcela Padilha Facetto. Azevedo, Young, S. B., and P. Setlow. 2004. “Mechanisms of Bacillus
Tamara Carolina de. Camargo, K U. Graziano, Subtilis Spore Resistance to and Killing by Aqueous
R A. Lacerda, and Ruth Natalia Teresa. Turrini. 2011. Ozone.” Journal of Applied Microbiology 96 (5):1133–42.
“Sterilization with Ozone in Health Care: An Integrative doi:10.1111/jam.2004.96.issue-5.