Implementation of Water Activity Testing to Replace Karl

www.pharmtech.com

WATER-ACTIVITY TESTING

58 Pharmaceutical Technology FEBRUARY 2007

indicates the total absence of water molecules. Water activ-ity is the effective mole fraction of water, defined as:

aw 5 lwxw = p/p0

in which lw is the activity coefficient of water, xw is the molefraction of water in the aqueous fraction, p is the partial pres-sure of water above the material, and p0 is the partial pres-sure of pure water at the same temperature (i.e., the wateractivity is equal to the equilibrium relative humidity [ERH]),expressed as a fraction. The relationship between water ac-tivity and weight percent water is shown in a sorptionisotherm (see Figure 1). Figure 1 also shows the hysteresisobserved when the sorption isotherm depends on whetherthe water is added to the dry material or removed from thewet material. This hysteresis is caused by nonreversible struc-tural changes and/or kinetic effects.

The sorption isotherm for a formulated drug product is theresult of the combination of the individual-component sorp-tion isotherms. Sorption isotherms for excipients and the drugsubstance may be used to model the sorption isotherm of thefinished dosage form (5). Figure 2 shows an example of anadsorption isotherm for a typical oral formulation. Two re-cently published studies present adsorption and desorptionisotherms for several common excipients (5–6).

Materials of differing water activity exhibit time- and temperature-dependent water migration from areas withhigh to low aw. For example, a gelatin capsule filled with apowder that has a different water activity will undergo watermigration and, at equilibrium, the capsule shell and powdercontents have the same aw. For packaged products, the moisture-vapor barrier properties of the packaging becomecrucial in the rate of moisture loss or gain. In addition, des-iccants inside a package change the initial water activity aswell as the subsequent rate of water activity change. Because

the humidity of the air is defined as 60% (aw = 0.6) by theInternational Conference on Harmonization under ClimaticZones II, materials with lower aw tend to gain water and thosewith higher aw tend to lose water. Most solid oral-dosageforms are produced with water activities of 0.3 to 0.5 andthus gain water in stability studies.

Use of water activity. The food industry has long used wateractivity to control microbial growth and chemical stability.Consistent with the US Food and Drug Administration’s useof water activity in controlling the microbial attributes offood (7), water activity also is being used to control the mi-crobiological properties of pharmaceutical products (8, 9).The growth of most bacteria is inhibited below approxi-mately aw 5 0.91. Similarly, most yeasts cease growing belowaw 5 0.87, and most molds stop growing below aw 5 0.80.The absolute limit of microbial growth is approximately aw 5 0.6.

The importance of water activity in controlling the micro-bial properties of pharmaceutical products was added to USPharmacopeia (USP) Chapter ^2023&, “Microbiological At-tributes of Nonsterile Nutritional and Dietary Supplements,”and a new USP chapter was written: ^1113& “Application ofWater Activity Determination to Nonsterile PharmaceuticalProducts.” USP Chapter ^1113& primarily describes the ap-plication of water-activity measurement for the control of anonsterile formulation’s microbial attributes, but also men-tions the control of chemical degradation. The correlation ofwater activity and chemical changes in food also has been in-vestigated extensively (10). A correspondingly common prac-tice in pharmaceutical development is to equilibrate prod-ucts at different humidities (water activities) and study theirdegradation rates. Water activity also correlates with changesin chemical and physical stability (11, 12). The degradationrate of several pharmaceutical products is well modeled bywater activity using the BET function (rate is proportionalto 1/[1 – aw]) (11) as well as other approaches (12–15).

Measurement of water activity.Water activity is determined bythe measurement of water in the gas phase immediately sur-

desorption

adsorption

aw0

0

Wat

er c

on

ten

t (%

)

0.5 1.0

50 100

No microbial growth ,0.6 aw

Stronglyboundmonolayer

Multilayer waterand capillary water

Solventand freewater

Relative humidity (%)

Figure 1: Water sorption isotherm.

Wei

gh

t ch

ang

e (%

)

Relative humidity (%)

8.0

7.0

6.0

5.0

4.0

3.0

2.0

1.0

0.00.00 20.00 40.00 60.00 80.00 100.00

Figure 2: Typical adsorption isotherm for tablet formulation.

www.pharmtech.com



rounding the sample. The measurement of water activity in foodsis described by AOAC’s Official Methods of Analysis, No. 978.18,“Water Activity of Canned Vegetables”(16). The principal tech-niques are changes in electrical capacitance or resistance of poly-mer thin films and dew point by chilled mirror. Newer tech-niques include the measurement of water in the headspace eitherspectroscopically (17) or chromatographically (18).

Method Number 978.18 also compares various types ofinstruments. For sensor-type instruments, a typical instru-ment chamber would have a depth of approximately 1.3 cmand a diameter of 4.2 cm. Sufficient sample is needed to fillthe chamber to the appropriate level (one-half to two-thirdsfull). Probes also are available for insertion into bulk mate-rials or possibly through product packaging. Measurementtime is controlled by sensor measurement time (usually a fewminutes) and time for the headspace to equilibrate (the timefor water activity in the headspace to come into equilibriumwith the material being tested). Typical equilibration timesvary from 5 to 30 minutes.

The rate of equilibration of the dosage form with the cham-ber’s headspace depends on several factors, including:• the difference in water activity between the dosage form

and the air in the chamber’s headspace immediately fol-lowing the sample-transfer process;

• the headspace volume of the measuring chamber contain-ing the sample;

• the transfer rates of water within the sample and betweenthe sample and the gas phase;

• the gas-phase diffusion rate for water in the headspace.The gas-phase diffusion rate of water is not a significant

factor based on the configuration of typical sensor cells. Thegas-phase diffusion rate is approximately 0.1 cm2/s and thelargest distance in the chamber is approximately 4 cm. Be-cause the sample occupies most of the chamber, the largestdistance is actually closer to 0.5 cm (from the top of the sam-ple to the top of the chamber). The transfer rate of water be-tween the sample and the gas phase is related to the specificsample for evaluation. Depending on the sample, the trans-fer rates within the sample and between the sample and thegas phase can vary significantly. Heterogeneous samples suchas intact capsules or coated tablets may have longer equili-bration times than uniform powder samples.

Although sample transfer can be performed rapidly to min-imize changes to the sample moisture content, moisture stillmust be absorbed or desorbed from the sample to equilibratethe measuring chamber’s headspace. For a typical sample, thechange in the sample’s moisture content during this equili-bration process will not have a significant effect on the accu-racy of the sample’s final aw reading. Air at a temperature of25 8C contains approximately 0.023 mg/mL of water at 100%RH. For a sample at 75% RH, with the external environmentat 0% RH and a sample chamber headspace of 6 mL, the sam-ple must replace only approximately 6 mL 3 0.023 mg/mL 3(0.75 – 0.10) or 0.01 mg of water. With a sample weight of 2g in the chamber and a water content of 2%, the total amountof water in the sample is 40 mg. In this case, only a very smallfraction of the total water is removed from the dosage formto equilibrate the humidity in the headspace (the water con-tent of the sample would drop by less than 0.01%).

Samples that contain very little water or have a very shal-low slope for the sorption isotherm in the region of interestwill be problematic with regards to bias during equilibrationin the chamber. The use of laser-absorption spectroscopy(which can determine the water activity in the headspace ofa sample vial without breaching the seal) would be beneficialfor these types of samples. An example of a product that couldbe difficult is a lyophilized cake with very low water content.Nonetheless, the typical capsule or tablet product generallycontains a large amount of water and is an excellent candi-date for water-activity testing.

For most solid oral-dosage forms, the range of interest forwater activity is 0.30–0.75. Water-activity measurements aregenerally precise and accurate to 0.01–0.02 units. Thus, wateractivity should be capable of providing 23 levels ([0.75 –0.30]/0.02) of distinction over this measurement range. Sev-

Table I: Examples of required changes in water activity todouble degradation rate.

Example

Change in aw todouble

degradation rate

Change in watercontent to doubledegradation rate

Cephalosporin degradation (19) 0.2 0.3–0.5%

N-oxide formation (20) 0.2 4.1–5.7%

Waterman aspirin degradation(21)

0.2–0.3(estimated)

Not available

Hydrolysis of nitrazepam (22) 0.1–0.2 Not available

Asparatamine degradation (23) 0.1–0.2 Not available

Rel

ativ

e h

um

idit

y (%

)

Temperature (°C)

80

70

60

50

40

30

20

10

015 20 25 30

NaCI (;75%)

Mg(NO3)

2 (;53%)

MgCI2 (;33%)

LiCI (;11%)

y 5 –0.0348x 1 76.148

y 5 –0.0298x 1 60.34

y 5 –0.0574x 1 34.189

y 5 –0.0014x 1 11.329

Figure 3: Relative humidity as a function of temperature forsaturated salt solutions used for calibration (only the magnesiumnitrate standard is sensitive to temperature).

www.pharmtech.com



eral examples of degradation as a func-tion of water activity are shown in TableI, and water activity changes of about0.1–0.2 are needed to double the degra-dation rate. In a case for which a 0.1change in water activity is critical, tightercontrols of calibration may be needed.In general, the water activity would becontrolled at a level well below when sig-nificant degradation occurs. Thus, theexamples in Table I represent mostly theworst cases in terms of required accu-racy. Compared with Karl Fischer watertesting, water activity has particular ad-vantages when:• the sample is hygroscopic;• the sorption isotherm is relatively flat

over the region where increaseddegradation rates are noted (i.e.,little change in water content);

• the amount of total water is high (e.g.,hydrates) and variability in the meas-urement makes discriminatingchanges in water content difficult.Calibration. Calibration is accom-

plished with the use of either saturatedsalt solutions or solutions with well-characterized water activities. Saturatedsalt solutions that have reference valuestraceable to National Institute of Stan-dards and Technology standards (24)are available. Figure 3 shows the effectof temperature on some typical water-activity standards.

ExperimentOver-the-counter tablets of ibuprofen,naproxen, and enteric-coated aspirinwere obtained from a local retail estab-lishment for evaluation. Two develop-mental capsule formulations also wereevaluated. Capsule A was a starch-basedformulation and Capsule B containedenteric-coated beads. The water-activityinstruments used were from the samemanufacturer (Novasina, Pfäffikon,Switzerland) but obtained from Omn-imark Instrument Corp. (Tempe, AZ).The names of the two models are theMs1 and the AW Lab.

Both instruments use the same meas-uring cell and sensor. The AW Labmodel has an additional capability ofdetermining whether the measurementhas reached equilibrium by establish-ing a limit of the change in water activ-ity with time. All measurements weremade using the mechanical prefilter, ex-cept for those made for Capsule B. Forthe Capsule B measurements, all of theprotective filters listed in Table II wereevaluated.

Water-activity standards (saturatedsalt solutions) were obtained from theinstrument supplier. Both Novasina in-struments take into account tempera-ture during calibration as long as thetemperature is between 15 8C and 30 8C.

The Novasina instruments do notallow for temperature control. To obtaintemperature-effect data, the measuringchamber was placed in a plastic bag andimmersed in a water bath. Measurementswere only recorded after the sensor tem-perature reading agreed with the exter-nal water-bath temperature.

Because it is not easy to predict theeffect of test samples on the instrumentsensor, bracketing readings of water ac-tivity standards were performed regu-larly (one low and one high standardreading during each period of use, typ-ically 24 hours).

The equilibration of dosage forms todifferent water activities were per-formed to ensure the validity of thewater-activity measurement. In general,the same type of salt solutions used insealed desiccators also were used for cal-ibration. Intact tablets or capsules were

Table II: Protective filters used withwater-activity instruments fromNovasina for Capsule Bmeasurements.

Commercialdescription

Function

Mechanicalprefilter/eVMT-2

Protects sensor fromparticles of sample

EVC-21Protects sensor fromacidic compounds

EVC-26 General chemical filter

EVALC-1Protects sensors fromalcohol

REDOXOxidizes, reducesinterfering compounds


www.pharmtech.com64 Pharmaceutical Technology FEBRUARY 2007

equilibrated for 7–14 days. Crushed orcomposited dosages were equilibratedfor 24 hours.

Studies requiring RH control wereperformed with a humidity-controlledglove box (Coy Laboratory Products,Grass Lake, MI).

Results and discussionKey operating conditions that were in-vestigated include the effects of sampletemperature, equilibration time, andsample handling (in particular, the hu-midity of the environment in whichsamples are transferred). In addition,the factors affecting the specificity ofthe sensor must be considered to sup-port the validation of the technique.Based on an earlier discussion, changes

in aw of .0.01 to 0.02 were not consid-ered significant.

Sample temperature.If the sample andthe equilibration chamber are at thesame temperature, then temperature hasa small effect on water activity withinthe temperature range of most labora-tories. The authors previously discussedthe effects of temperature on the wateractivity of the saturated salt standardsolutions used for calibration. Those ef-fects are generally small and taken intoaccount by the instrument. The effectof temperature on solid oral-dosageforms was studied for two samples: Cap-sule A and naproxen. The observed ef-fects of temperature changes on wateractivity were 0.003 and 0.004 aw /8C andwere similar to those reported by a man-

Wat

er a

ctiv

ity

Time (min)

0.45

0.44

0.43

0.42

0.41

0.4

0.39

0.380 200 400 600 800 1000 1200 1400

Time (min)

0.45

0.44

0.43

0.42

0.41

0.4

0.39

0.380 10 20 30

Wat

er a

ctiv

ity

Figure 4: Rate of equilibration of Capsule A. W

ater

act

ivit

y

Time (min)

0.49

0.47

0.45

0.43

0.41

0.39

0.37

0.35

0.330 50 100 150 200 250 300 350 400

Naproxen

Enteric-coatedaspirin

Ibuprofen

Figure 5: Rate of equilibration of several solid oral-dosage forms.

Pharmaceutical Technology FEBRUARY 2007 6565 Pharmaceutical Technology FEBRUARY 2007

ufacturer (25). The small effect of tem-perature is explained by the fact that thechange in the partial pressure of watervapor (p) with temperature for the sam-ple is similar (but not equal) to the mag-nitude of the change of the saturationpressure (p0) above pure water. Anychange in the temperature of a hygro-scopic product automatically causes theproduct to exchange moisture with theair (or gas) that surrounds it. Moistureis exchanged until the partial water-vapor pressure at the product’s surfaceand in the air is equal. For this reason,a constant temperature is importantwhen measuring aw.

Equilibration time. To understand thetime required for the headspace to comeinto equilibrium with the sample, wateractivity was measured as a function oftime for the products shown in Figures4 and 5. The results for Capsule A indi-cated that equilibration occurred within10 minutes and no significant changesoccurred over the subsequent 12 hours(see Figure 4). Figure 5, which includestwo typical tablet formulations and anenteric-coated aspirin tablet, shows thata 30-minute equilibration time was ap-propriate for the tablet formulations. Alonger equilibration time would beneeded with the enteric-coated formu-lation for the same degree of accuracy.In general, the authors have used an ac-curacy of 60.02 aw as being acceptable.Another alternative would be to crushthe enteric-coated tablet, but this pro-cedure needs special precautions.

For some instruments, the equilibra-tion time is not a concern. These instru-ments measure the rate of change of aw

with time and will indicate a stablereading only when the rate of change isbelow a selected threshold. Other in-struments use a mathematical functionto estimate the final equilibrium value.The advantage of measuring water ac-tivity on intact tablets is shown in TableIII, in which the water activity of intacttablets is compared with that of crushedtablets under 10% and 80% RH. Thetotal exposure time to the ambient RHwas approximately 2 min (including thetime to crush the tablets with a mortarand pestle). This time was used to sim-

Table III: Effect of sample grinding on measured water activity.

Experiment Water activityError compared with

intact

Intact ibuprofen tablets (different lot thanin Figure 5)

0.345 —

Ibuprofen tablets crushed in a 80%relative humidity environment (2 minexposure)

0.530 153%

Ibuprofen tablets crushed in a 10%relative humidity environment (2 minexposure)

0.321 –7%

www.pharmtech.com



ulate rapidly crushing, weighing, and transferring tablets inalternative methods such as Karl Fischer. Whereas the lowhumidity only created a small bias, the high humidity cre-ated a large error. There was a larger difference between thesample water activity and the ambient conditions for thehigh humidity (45% RH higher) versus the low humidity(22% RH lower). The rate of water uptake versus loss is notexpected to be the same. A technique for reducing waterchanges during crushing is to use a device sold to home cus-tomers that sandwiches the sample between two holders,thus reducing contact with the ambient atmosphere.

Sample handling. The effect of ambient humidity on sam-ple handling was assessed by transferring Capsule A in en-vironments of 10% and 80% RH. As shown in Figure 6, thee

www.pharmtech.com


ated with an enteric-coated aspirin tablet and two other tabletformulations as shown in Table IV with a general accuracyof 0.01–0.02 aw. The same approach was taken by Mahajanet al., with a similar degree of accuracy (17). As previouslymentioned, a 30-minute equilibration time generally hasbeen adequate for capsules and tablets.

Precision.Replicate measurements were made for two tabletformulations and one capsule formulation (see Table V). Therelative standard deviations for replicate measurements were1–3%.

Instrument qualification of maintenance (Novasina Instrument).Using the guidelines published in a recent white paper (27),water-activity instruments would be Class B, principally re-q

70 Pharmaceutical Technology FEBRUARY 2007 www.pharmtech.com


• validating methods once they are being used for lot-releasepurposes.Table VI describes some differences in the types of instru-

ments. Only representative manufacturers are included.The approach used by all of the instruments is nondestruc-

tive. The instruments vary in the cutoff at low RH, the con-trol of temperature, and the detection of the equilibrationend-point. Measurement time is faster for the laser absorp-tion and the instruments that extrapolate the detection of theend-point; but, temperature control generally lengthens themeasurement time to 10–30-minutes for all the instruments.The Lighthouse instrument would be especially useful forlyophilized formulations because the measurement can bemade on the headspace of the product vial. Reference 16 hasmore details about differences in cost and applicability.

The control strategy also must be consistent with the reg-ulatory strategy. Table VII summarizes options relative to theproduct registration status.

Routine control of instrumentation. For the instruments thathave sample contact with the sensor, bracketing readings ofthe standard solutions should be used to ensure that the sen-sor is providing accurate results and that the samples did notcontaminate the sensor. The use of a protective filter shouldbe tracked such that the filter is replaced at least as frequentlyas the manufacturer suggests. Standards that are saturated

salt suspensions also require some maintenance to replacelost water from evaporation. These standards should bechecked each time they are used.

Table VII: Options for implementing water-activitytesting as a function of registration status.

Product registration status Options

Marketed product withKarl Fischer water test

Establish water activity as analternate procedure with a limit basedon sorption isotherm.

If actual water levels are close toregistration limit, then variations inisotherms should be considered.

Marketed product withno registered KarlFischer water test

Follow normal practices for changinganalytical procedures.

Product startingregistration stability orearlier

1. Implement with the intent to usewater activity as an internal GMPtest (e.g., not registered).

2. Correlate Karl Fischer and wateractivity tests to keep options open,but use water activity as theprincipal test.

Pharmaceutical Technology FEBRUARY 2007 7171 Pharmaceutical Technology FEBRUARY 2007

ConclusionsWater activity is a valid alternative to total water measure-ments (KF) in assessing the potential for water to adverselyaffect the formulation’s microbial or chemical quality. Wateractivity has been accepted in the food industry and widelydiscussed in the pharmaceutical industry as an appropriatemeasurement for water to prevent microbial growth. Like-wise, water activity also can be used to evaluate the poten-tial effect of water on chemical and physical changes in theformulation. The Novasina ms1 and AW instruments in par-ticular have proven to be sufficiently precise and accurate tomeasure the water activity of typical solid oral-dosage forms.

AcknowledgmentsThanks to Jim Farmer for his assistance.

References1. M. Mathlouthi, “Water Content, Water Activity, Water Structure

and the Stability of Foodstuffs,” Food Control 12 (7), 409–417 (2001).2. C. Ahlneck and G. Zografi, “The Molecular Basis of Moisture Ef-

fects on the Physical and Chemical Stability of Drugs in the SolidState,” Int. J. Pharm. 62 (2–3), 87–95 (1990).

3. S. Brunauer, P.H. Emmett, and E. Teller, “Adsorption of Gases inMultimolecular Layers,” J. Amer. Chem. Soc. 60, 309–319 (1938).

4. J. Maques

Documents

Implementation of Water Activity Testing to Replace Karl