Medical Device & Diagnostic Industry Magazine
MDDI Article Index

Originally Published MDDI October 2005

Sterilization

With the possibility of parametric release, low-temperature gas plasma
sterilization may be a viable alternative to ethylene oxide for processing
heat-sensitive devices.

By Benjamin M. Fryer and James P. Kohler

Approximately 15 million sterilization cycles of heat-sensitive medical devices are performed in U.S. hospitals every year.^1,2 Sterilization of heat-sensitive devices was once dominated by EtO, but this technology has recently been shown to have health and regulatory concerns related to EtO toxicity.³ An increasingly popular alternative to EtO for sterilizing heat-sensitive medical instruments is low-temperature gas plasma (LTGP). LTGP sterilization is now performed approximately 5.5 million times a year in healthcare settings worldwide.⁴ In this process, an aqueous hydrogen peroxide solution boils in a heated vaporizer and then flows as a vapor into a sterilization chamber containing a load of instruments at low pressure and low temperature. Exposure to the hydrogen peroxide vapor and plasma for a controlled time completes the sterilization of medical instruments.

LTGP sterilization cycles are monitored with chemical and biological indicators within the instrument load. A biological indicator (BI) is "a device consisting of a known number of microorganisms, of known resistance to the mode of sterilization, in or on a carrier and enclosed in a protective package."⁵ After sterilization, the BIs are cultured to determine whether the cycle was sufficient to sterilize the load by disabling 100% of the microorganisms.

Several problems are associated with using BIs to monitor a sterilization cycle. Problems include the time required for BI incubation, the occurrence of false positives, correct handling of the BIs, and the requirement for expensive microbiological assay equipment. A method to monitor LTGP sterilized loads parametrically, by measuring the critical sterilization parameters related to hydrogen peroxide exposure, would save time and would be less labor-intensive. Well-established sterilization methods, such as steam and EtO, have procedures for verifying parametric release, but no accepted methodology currently exists for LTGP sterilization.^3,6-8

The newest version of LTGP technology concentrates hydrogen peroxide by removing water from the solution before introducing it as a vapor into the sterilization chamber. As in earlier LTGP processes, the important variables for sterilization are sterilizer temperature, process pressure, exposure time, and hydrogen peroxide concentration. Only the first three parameters are monitored during the traditional LTGP processes, but the latest LTGP technology platform also monitors the peroxide concentration in the sterilization chamber during the peroxide vapor exposure step. With a complete set of monitored variables, it is now possible to establish a parametric release methodology for this sterilization technology.

This article describes the theoretical mechanism for delivery of hydrogen peroxide vapor into the chamber. It also looks at a hydrogen peroxide concentration-time integration method for determining the lethality of LTGP cycles on microorganisms.

Theory of Operation

In the newest LTGP process (see Figure 1), hydrogen peroxide and water boil in a vaporizer during the vaporization step. The vapor then flows through a condenser, where hydrogen peroxide collects as a concentrated solution and water vapor is discharged from the sterilizer. In the following transfer step, hydrogen peroxide boils in the condenser and the resulting vapor flows into the chamber containing the medical instruments. The flow of hydrogen peroxide vapor from the condenser into the chamber is governed by a partial differential equation:⁹

ðc_p/ðt + . c_pv = c D (c_p/c) + r_p, (1)

where c_p = hydrogen peroxide mass concentration at the point of interest in the chamber, in g/cm3,
t = time, in seconds,
= differential operator, ð/ðx + ð/ðy + ð/ðz, a vector function, in cm^–1,
v = mass average velocity for the vapor phase, a vector quantity, in cm/sec,
c = mass density of the vapor phase, in g/cm3,
D = diffusion coefficient for hydrogen peroxide, in cm²/sec, and
r_p = reaction term, g/(cm³ . sec).

Equation 1, the species continuity equation, is a mass balance on hydrogen peroxide vapor at each point in the chamber within an arbitrary region where the mass balance is measured. It states the following: The time rate of change of hydrogen peroxide mass per volume at a point inside the chamber within the mass balance region, plus the net mass per volume entering the region around the point per time by convection, equals the net mass per volume entering the region per time by diffusion, plus the net mass of hydrogen peroxide per volume per time produced within the region.

If hydrogen peroxide disappears from the vapor phase inside the chamber by absorption, adsorption, chemical decomposition, or condensation, the reaction term is negative and represents a net loss in the equation. When the vapor is uniformly distributed within the mass balance region, Equation 1 simplifies because the vapor density c and the diffusion coefficient D are invariable with location:

ðc_p/ðt + v . c_p = D ²c_p + r_p, (2)

where ² = differential operator, ð²/ðx² + ð²/ðy² + ð²/ðz², in cm^–2.

If the mass balance region in the chamber is chosen to be the narrow zone where an ultraviolet light sensor detects hydrogen peroxide vapor, the flow into the chamber may be treated as a one-dimensional process to simplify the model. The equation in the x-direction becomes

ðc_p/ðt + v_x ðc_p/ðx = D ð²cp/ðx² + r_p. (3)

The convection term on the left side of Equation 3 requires the mass average velocity of the vapor in the x direction v_x, which is obtained from the equations of motion, a momentum balance on the vapor in the chamber. Convection into the chamber is driven by the pressure gradient between the condenser and the chamber. Chamber pressure increases rapidly during the first few seconds of transfer of hydrogen peroxide and rises slowly for the remainder of the transfer step. Hydrogen peroxide flow into the chamber by convection, therefore, is greatest in the first few seconds of the transfer step.

The diffusion term on the right side of the equation depends on the diffusion coefficient and the gradient of hydrogen peroxide concentration in the region of interest, the zone near the hydrogen peroxide sensor.

This form of the equation can be solved numerically with detailed knowledge about the velocity and concentration profiles and the reaction term. However, this information is difficult to obtain during a sterilization cycle. Therefore, experimental evidence is used in the results section below to estimate the relative magnitudes of the convection, diffusion, and reaction terms in the equation.

Materials and Methods

Spore Production. Spores for inoculation were prepared by suspending the lyophilized seed culture of Geobacillus stearothermophilus (ATCC #7953) in sterile DI water. Aliquots (0.2 ml) of this suspension were inoculated onto trypticase soy agar (TSA) plates, spread with a glass L-shaped rod, and incubated at 55°C for 2–5 days. The colonies were removed from the TSA plates with 2 ml of WFI (sterile water for irrigation USP, Abbott Labs) and a glass L-shaped rod. The resulting solution was centrifuged at 2350 g at 4°C for 25 minutes. The pellet was then washed with sterile WFI and filtered through a 60-mesh funnel. The resulting suspension was sonicated for 3 minutes. The centrifuging and washing steps were repeated until the supernatant was colorless to the eye. After the final washing, the pellet was suspended in sterile WFI and stored at 4°C until use.

Preparation and Inoculation of the Biological Indicators. The substrates used were the keyway section of a stainless-steel surgical blade (Bard-Parker) and 0.015-in.-diameter stainless-steel wire (Fairbanks Wire Corp.). The keyway coupon was approximately 2 × 10 mm. The degreased wire was cut to lengths of approximately 60 mm. Once dried and cooled, approximately 106 spores in the inoculum were pipetted onto each substrate.

Tubing. The tubing used included 1- and 2-mm-diameter stainless-steel tubes (Accu-Tube Corp.). These were cut to 150-, 400-, and 500-mm lengths, and then degreased and washed in a pipette washer. Also used were 1-mm-diameter polytetrafluoroethylene (PTFE) and polyethylene (PE) tubes (Becton Dickinson). These were cut to 350-mm lengths.

Hydrogen Peroxide Solution. The hydrogen peroxide solutions used were 53 and 59 wt% (Univar).

Test Load. The tubes with BIs were placed into an instrument tray with a selection of stainless-steel and plastic instruments on a silicone mat (see Figure 2) to simulate a heavy hospital sterilization load. The tray was then double wrapped in KimGuard KC400 (Kimberly-Clark).

Operation. A Sterrad NX sterilizer was used to perform all spore inactivation experiments. The process is based on using vapor-phase hydrogen peroxide to inactivate microorganisms and a 50-kHz discharge plasma to reduce hydrogen peroxide residuals in the load. The series of process steps is shown in the diagram in Figure 1.

The sterilizer used in situ vaporization to concentrate the initial 53 wt% solution before the transfer stage occurred. For vapor-phase hydrogen peroxide detection, UV absorption measurements were taken at 254 nm. The length of hydrogen peroxide exposure was 3 and 7 minutes for the standard and advanced cycles, respectively.

Test Method. The test loads were placed in the sterilizer chamber and the process cycle started. The BIs were exposed to the half standard and advanced cycles with the hydrogen peroxide injections shown in Table I. A sample of 10 BIs was used for the most probable number (MPN) tests,¹⁰ and three BIs were used for each plate count.

Surviving Spores. At the end of the process, the test loads were removed from the chamber and the number of surviving spores on each processed BI was determined by either MPN or plate counts. For the plate counts, the BIs were aseptically transferred to screw-top tubes with glass beads, vortexed for 2 minutes, chilled in an ice bath for 2 minutes, vortexed for 30 seconds, and then serially diluted. The dilutions were plated on TSA and incubated at 55°C for 2 days. The number of colonies was counted manually. For the MPN method, the BIs were aseptically transferred into trypticase soy broth (Becton-Dickinson), incubated at 55°C, and checked for turbidity daily for 2 weeks.

Calibration of the Hydrogen Peroxide Monitor. The hydrogen peroxide monitor was calibrated by vaporizing known weights of hydrogen peroxide solution into the sterilization chamber. The chamber was heated to approximately 55°C to eliminate hydrogen peroxide condensation. Figure 3 shows the resulting calibration curve, a cubic regression of the data and the 95% confidence interval (CI). The regression has an adjusted r² of >99%. The standard error of the absorbance mean (a = 0.05%) of 0.0046 corresponds to an average 0.34 mg/L of error around the mean using this regression equation. Regression and confidence intervals were calculated with Minitab version 14.1.

Area under the Hydrogen Peroxide Concentration versus Time Curve (AUC) Calculation. The hydrogen peroxide concentration was integrated by summing the measured concentration each second. The calculations were done during the transfer stage when there was a >0.1-mg/L hydrogen peroxide concentration in the sterilization chamber.

Results and Discussion

Estimating the Magnitude of the Reaction Term in the Flow Equation. Medical instruments in the load and sterilizer components inside the chamber, such as the electrode, shelf, or supports, complicate the analysis of the terms in Equation 3. These items contribute to absorption, adsorption, condensation, and decomposition. However, studies with hydrogen peroxide solution injections into a completely empty sterilizer chamber show that the reaction term is small over a short time span, compared with the convection and diffusion terms. Figure 4 plots the hydrogen peroxide vapor concentration and pressure versus time in the sterilizer chamber at an injection volume of 0.4 ml of 59 wt% solution (0.5 g), a volume that is small enough to avoid condensation in the chamber. Concentration of the solution occurs in the condenser during the vaporization step of the process and reduces the solution mass.

The concentration and pressure rise rapidly in the first 10 seconds of the transfer step, as the vapor flows to the chamber. After the hydrogen peroxide distributes evenly throughout the chamber, the flow terms in the equation may be neglected, and only the transient term and reaction term remain. The slow decrease in concentration of about 0.5 mg/L during the last 300 seconds demonstrates the relatively small effect of the reaction term during the first few seconds of hydrogen peroxide flow to the chamber. In the case of an empty chamber, Equation 3 simplifies when the reaction term is deleted.

ðc_p/ðt + v_x ðc_p/ðx = D ð²c_p/ðx². (4)

Estimating the Magnitudes of the Diffusion and Convection Terms. In EtO sterilization, diffusion is commonly assumed to be the dominant mechanism of sterilant delivery to the load.¹¹ By contrast, as described below, LTGP process dynamics during the transfer step are dominated by convection. This result is obtained by calculating each term in the flow equation separately to assess its contribution.

The maximum value of the diffusion term may be estimated by neglecting the convective term and by assuming that diffusion of hydrogen peroxide from the condenser to the chamber is analogous to one-dimensional heat transfer in an infinite slab of material. Equation 4, with initial and boundary conditions, simplifies to

ðc_p/ðt = D ð²c_p/ðx². (5)

Initial Condition: At t = 0, the hydrogen peroxide concentration c_p = c₀ for x = 0 at the surface of the slab (in the condenser), and c_p = 0 everywhere inside the slab (in the path to the chamber).

Boundary condition: For t > 0, the hydrogen peroxide concentration c_p = c₀ for x = 0 at the surface of the slab (in the condenser).

The solution to this diffusion problem in one dimension as presented by Schneider is

c_p = c_o{1–erf[x/(2(Dt)^1/2)]}, (6)

where erf is the error function.¹²

The concentration of hydrogen peroxide in the empty chamber is calculated by the diffusion model in Table II with a correlation for the diffusion coefficient derived by Slattery and Bird, D = 3.303 × 10^-4 x (T₀K)^2.334/P; where x = the 46-cm diffusion path length from the condenser to the chamber.¹³ This calculation assumes that the pressure P in the condenser is constant at 10 Torr.

In Table II, the calculated hydrogen peroxide concentration increases from 0 to 1 mg/L during the first 10 seconds of transfer. By contrast, the observed concentration in the chamber from Figure 4 increases to 4.2 mg/L during this time. The plot of calculated and observed concentrations together as a function of transfer time (see Figure 5) clearly shows that the flow rate estimated by the diffusion model is much slower than the observed rate. This result suggests that diffusion is not the dominant mechanism for mass transport into the sterilizer chamber. The maximum value of the convective term in Equation 4 is estimated by neglecting the diffusive term:

ðc_p/ðt + v_x ðc_p/ðx = 0. (7)

The diffusion model assumes that the pressure in the condenser during the transfer step is constant at 10 Torr in order to calculate the maximum diffusion flow rate. In fact, the condenser pressure decreases during the transfer step, as shown in Table III, while the chamber pressure increases as hydrogen peroxide vapor enters the chamber.

The convective flow term, therefore, is largest early in the transfer step and smallest at the end, when the pressure gradient is least. The vapor flows from the condenser through an orifice to the chamber. Flow through the orifice is sonic during the first 3 seconds of the transfer step (see Table IV), because the pressure ratio is less than the critical pressure ratio r for sonic flow:⁹

P_chamber/P_condenser at 3 seconds = 1.7 mmHg/3.8 mmHg = 0.45 < r = (P_downstream/P_upstream)_critical = [2/(g + 1)](^g/g-1) = 0.54,

where g is the ratio of heat capacity at constant pressure to heat capacity at constant volume, and P is the pressure in the condenser (upstream) or chamber (downstream). The heat capacity ratio g for hydrogen peroxide vapor is approximately 1.3. In sonic flow, the transient term in the equation drops out, and the maximum mass flow rate is determined for the steady-state case to be

W_max = S_orifice {g P_vaporizerr_vaporizer [2/(g + 1)]^(g+1/g-1)}^0.5, (8)

where W_max = maximum mass flow rate in sonic flow, S_orifice = cross-sectional area of the orifice in the inlet valve, and r_vaporizer = density of the vapor in the condenser.⁹

Transfer Time (sec) Condenser Pressure (mmHg) Chamber Pressure (mmHg) Chamber Pressure/Condenser Pressure 0 10.0 0.3 0.03 1 5.4 1.0 0.18 2 4.5 1.4 0.31 3 3.8 1.7 0.45 4 3.4 2.0 0.59 5 3.1 2.2 0.71 6 3.0 2.4 0.60 7 2.9 2.5 0.68 8 2.9 2.7 0.93 9 2.9 2.7 0.93 10 2.9 2.8 0.98 Table III. Convection model: critical flow.

Table IV shows that W_max is about 0.1 g/sec. Therefore, the model predicts that the vapor should flow to the chamber within about 2.5 seconds to reach the maximum hydrogen peroxide concentration of 4.7 mg/L. However, the observed hydrogen peroxide concentration in the chamber after this time is only about 1.9 mg/L (see Figure 5). This result suggests that convective flow dominates the process, but that another factor is limiting the flow rate.

Most of the heat for vaporizing the concentrated hydrogen peroxide in the first few seconds, about 50 calories, comes from the sensible heat stored in the condenser and from the heaters. The rest is the sensible heat of the liquid itself.

Transfer Time (sec) Convection Model Flow Rate W_max (g/sec) Convection Model Mass in Chamber (g) Convection Model Hydrogen Peroxide Vapor Concentration in Chamber (mg/L) Observed Hydrogen Peroxide Vapor Concentration in Chamber (mg/L) 1 0.12 0.12 2.2 0.9 2 0.10 0.22 4.0 1.5 3 0.068 0.31 5.5 2.2 Table IV. Convection model: Convection model and observed flows.

Heat is transferred from the heaters to the aluminum condenser surface and then to the liquid drop. As droplets of liquid dance over the hot surface of the condenser, a vapor film forms between the droplets and the hot surface. The vapor film reduces the heat transfer coefficient, so heat is transferred less efficiently from the condenser surface to the drop. As a result, the flow of vapor to the chamber is controlled by the efficiency of heat delivery to the liquid phase in the condenser.

Figure 5. Hydrogen peroxide chamber concentration: diffusion and convection models versus observed.

An estimate of the relative magnitude of the conductance for heat transfer in the condenser is available from the dimensionless heat-transfer quantity, the Nusselt (Nu) number

Nu = hd/k,

where h = vapor film heat transfer coefficient, d = vapor film thickness, and k = thermal conductivity of the condenser.

Bird estimates that the heat-transfer coefficient for boiling water in free convection is a maximum of about 4000 Btu/(hr°Fft²).⁹ The thermal conductivity of aluminum is about 120 Btu/(hr°Fft²). For a 1-mm-diameter liquid droplet in the condenser, a reasonable vapor film thickness would be approximately 0.1 mm. With these assumptions, the Nusselt number becomes 0.01. The Nusselt number shows that the conductance for heat transfer in the aluminum condenser is 100 times greater than that in the vapor film beneath the liquid droplets. This result supports the assertion that heat transfer through the vapor film controls the vaporization process.

A clearer understanding of the mechanism for delivery of hydrogen peroxide to the chamber now opens the way to quantifying the exposure of the instrument load to hydrogen peroxide. BIs prepared with spores resistant to hydrogen peroxide were used to correlate efficacy with the parametric index of the process.

Hydrogen Peroxide Vapor Concentration. Figure 6 presents hydrogen peroxide concentration curves for the representative injection volumes used in this study. These data show typical hydrogen peroxide concentration profiles in a sterilizer chamber with a 23°C load. The worst-case condition for efficacy is 18°C; therefore, some cycles were run with loads at that temperature. The normal condition for customer loads is 23°C, so the hydrogen peroxide curves were measured with loads at that temperature.

Figure 6. Typical hydrogen peroxide profiles for representative injection volumes.

The shape of the hydrogen peroxide vapor concentration curves over time displays the convection of vapor into the chamber from the condenser and the distribution into the load, as discussed above. The rapid increase in concentration at the beginning of the transfer step occurs as the hydrogen peroxide vapor flows down the condenser-chamber pressure gradient.

As the hydrogen peroxide distributes throughout the chamber and interacts with the load and chamber internal components, the concentration decreases due to several factors. These include absorption by the chamber and its components, adsorption on the load wrapping, condensation on the instruments, and decomposition by chemically active materials in the load. Greater injection volumes produce greater peak concentrations in the first few seconds of transfer and greater final concentrations after losses from the vapor phase.

Evaporation of condensed hydrogen peroxide from the instruments would occur as the load temperature increased with longer time in the chamber. However, this mechanism does not influence the process dynamics because the exposure time to hydrogen peroxide is short (3 and 7 minutes for the standard and advanced cycles, respectively).

Sterilization Kinetic Model

Tests with incrementally increasing hydrogen peroxide injection volumes were conducted with 1 × 150-mm and 2 × 400-mm stainless-steel and 1 × 350-mm PTFE and PE tubes with standard cycle conditions. Advanced cycle conditions were tested with 1 × 500-mm stainless-steel tubes. The number of surviving spores on each BI was determined either with the MPN or with the direct-culture method, depending on the injection volume. Range-finding studies (not shown) determined which culture method was appropriate at each injection volume. The milligram-per-liter-per-second integral was calculated from hydrogen peroxide monitor data gathered during the cycle. Figure 7 shows the log of remaining spores counted at different injection volumes for 18°C loads, standard and advanced cycles.

Figure 7. Scatter plot of the log of spores surviving in standard and advanced cycles versus AUC.

In this study, the kinetics of spore deactivation are related to the two most important process variables in low-temperature gas plasma sterilization, hydrogen peroxide concentration and time. The AUC represents a convenient index for quantifying the exposure of medical instruments in the load to these variables. AUC values are calculated by summing the hydrogen peroxide concentration each second. In Figure 7, the surviving spores decrease in a nearly log-linear function versus AUC in both the standard and advanced cycles.

The AUC required for complete inactivation of BIs in 1 × 150-mm and 2 × 400-mm stainless-steel and 1 × 350-mm PTFE and polyethylene tubes with standard cycle conditions is approximately 250 mg/L/sec. The corresponding value with the advanced cycle in 1 × 500-mm stainless-steel tubes is 600 mg/L/sec. Inclusion of a monitor error of 61 and 142 mg/L/sec produces a target AUC of 311 and 742 mg/L/sec for the standard and advanced cycles, respectively.

These values apply only to this sterilization process and particular process variables in this sterilizer model. They would not be valid for other low-temperature gas plasma processes. Ambient conditions of air temperature and humidity have a negligible effect on the process because air is removed from the chamber before the hydrogen peroxide transfer step.

A greater AUC is required to completely inactivate spores in the advanced cycle configuration, because 1 × 500-mm stainless-steel tubes are more flow-restricted than the shorter tubing. The longer tubes in the advanced cycles present more surface area for the hydrogen peroxide vapor to contact, so less hydrogen peroxide remains in the vapor phase inside the tube to inactivate the spores on the BI. Therefore, a longer exposure time is required in the longer tubes to achieve inactivation, which results in a greater AUC when concentration is integrated over time.

Conclusion

The complex equation for flow of hydrogen peroxide from the condenser to the chamber reduces to a simple form with assumptions about the transfer of heat, chemical reaction, and flow behavior. Convection appears to dominate diffusion as the likely mechanism for flow of hydrogen peroxide into the chamber. The theoretical maximum vapor flow rate into the chamber is not attained, because the vapor film surrounding each droplet of liquid limits the rate of heat transfer to the liquid in the condenser. The spore inactivation kinetics in a hydrogen peroxide gas plasma sterilizer can be described by integration of the hydrogen peroxide monitor curves. This model can be used as part of the criteria for parametric release of medical devices sterilized in hydrogen peroxide gas plasma sterilizers.

Acknowledgment

The authors would like to acknowledge Dan Smith, Martin Favero, and Szu-Min Lin for their review of this manuscript and for the data provided by Veronica Thralls and Terry Rhee.

References

1. Sylvia Sfiligoi, personal communication.
2. C Werner, "Sterilization Market a Vibrant, Active Segment," Health Care Purchasing News 27, no. 2 (2003): 40–42.
3. KJ Hemmerich, J Masefield, JR Nelson, "Sterilization Methods Stand the Test of Time," Medical Device & Diagnostic Industry 26, no. 8 (2004): 132–145.
4. David Fraser, personal communication.
5. 21 Code of Federal Regulations 880 2800 (a) (1).
6. ANSI/AAMI/ISO 11135 (Washington, DC: American National Standards Institute, 1994).
7. Irving J Pflug, ed., Selected Papers on the Microbiology and Engineering of Sterilization Processes, 5th ed. (Minneapolis: Environmental Sterilization Laboratory, 1988).
8. Gregg A Mosley, John R Gillis, James E Whitbourne, "Formulae for Calculations of Integrated Lethality for EtO Sterilization Processes," Pharmaceutical Technology 26, no. 10 (2002): 114–136.
9. RB Bird, WE Stewart, and EN Lightfoot, Transport Phenomena (New York: John Wiley, 1962), 393, 482, 557.
10. Seymour S Block, ed., Disinfection, Sterilization and Preservation, 4th ed. (Media, PA: Williams and Wilkins, 1991), 118–120.
11. Z Zhu, IP Matthews, "Process Dynamics of EtO Sterilization," Medical Device & Diagnostic Industry 25, no. 3 (2003): 66 –80.
12. Paul J Schneider, Conduction Heat Transfer (Cambridge, MA: Addison-Wesley, 1955).
13. JC Slattery and RB Bird, "Calculation of the Diffusion Coefficient of Dilute Gases and of the Self-Diffusion Coefficient of Dense Gases,"American Institute of Chemical Engineers Journal 4, no. 2 (1958): 137–142.

Copyright ©2005 Medical Device & Diagnostic Industry