Saturday, 7 October 2006

INTC, Houston , Sept 25th – 28th 2006

Key Points

• Peptides have biocidal activity against bacteria and fungi in microgram quantities and are safe for the environment.
• Antimicrobial filter layers have been made by electrospinning inherently antimicrobial polymers.
• Copper containing nonwoven fabrics possess potent antibacterial, antifungal and antiviral properties which are unaffected by washing.
• The simulation of nonwoven geometry, the fluid-flow associated with that geometry, and particle deposition on the fibers, makes feasible the virtual design of a filter.
• The deviation of the surface coefficient of friction correlated best with the subjective assessment of handle of diaper and pantiliner topsheets.
• Nylon filaments with a diameter of 360 nm are liberated when a 360 island/sea fiber with a 25/75 nylon/PLA ratio has the PLA sea removed. They have high tenacity (5 gf/denier) and modulus (140 gf/den) and nonwovens made from them have very high tear strength.
• Nordson's bico melt-blow process is giving webs with an average fiber diameter of 0.75 micron, yielding double the usual hydrohead.

  This years INTC, excluding the student papers and the tutorial session, comprised 74 papers in 15 separate sessions arranged in 3 concurrent streams. Jointly organised by INDA and TAPPI, this meeting had a “research academy” feel - like the old TAPPI conferences - 42 of the papers being from universities or research institutes. Only the 25 papers which could be attended are summarised here.

Charged Nonwoven Materials

Kyung-Ju Choi, AAF International said the point of electrically charging filters was to achieve higher efficiency without increasing air resistance. The common ways of charging were:

• Corona discharge
• Triboelectric charging
• Induction charging (as in electrospinning)

He has characterised the structure and thermal properties of charged nonwovens from homo and hetero-polymers using optical microscopy, scanning electron microscopy (SEM) and differential scanning calorimetry (DSC). The charge distribution was checked with an electrostatic fieldmeter and filtration efficiency and resistance measured.

He concluded:

• Fiber orientation was random for dry-laid, wet-laid and spunbond nonwovens except for non-crosslapped dry lay.
• Fiber sizes ranged from 9.5 to 40 micrometers for dry-laid and wet-laid nonwovens.
• The residual static voltages on a polypropylene meltblown web ranged from -0.38 to +0.53KV.
• These static voltages moved around slowly with slight changes of maximum and minimum voltages.
• The filtration efficiency of polypropylene meltblown nonwovens increased with increasing voltage applied.
• The filtration efficiency of charged polyester spunbond, dry-laid, and wet-laid materials doubled c.f. the controls but the efficiency dropped completely in 3 days.
• PP spunbond loses charge during the first day but is then stable.
• Electrospun PTFE holds its charge for a month.

Can the microstructure of charges be measured? Not yet – he's working on it.

Feather Fiber in Filters

Roy Broughton replaced Xiuling Fan of Auburn University to present this paper. The USA produces 2-3 billion lbs (dry weight) of dirty chicken feathers annually and these are usually fed to chickens because they are not easily disposed of in other ways. However the US government may follow the EU and ban this practice so funds are available to find other uses for these feathers. Possibilities include insulation, reinforcement, papermaking and protein fiber production. AU is cleaning them, stripping out the coarse backbone and air-laying a polyester blend of the small branch feathers on a scrim prior to needlepunching using 4 passes to give a total of 1000 punches per inch. Fabric physical and filtration properties were measured.

Although feather fiber could be recycled into air-filter fabrics, its fineness and the tree/fan-like structure of the feather did not offer a high level of performance advantages over conventional fibers. The use of feather fiber in air filtration applications must rely primarily on a favorable cost and weight differential in favor of the feather fiber, assuming there is one after the cleaning and dequilling process. Prof Broughton alluded to a chicken and egg situation. No one was prepared to clean up the feathers until a market had been found, and vice versa. Cleaning would yield about 40% of nonwoven-processable feather, so there would still be nearly a couple of billion pounds of waste and its dirty water to dispose of.

Aerosol Filter Tests

Sven Schütz, Palas GmbH (Germany ) reviewed round robin testing of car cabin air filters in 6 labs in the USA and Europe during 1999-2004 using ISO test methods (A2 Dust Test and DIS111155-1 KCl test). Electreted and electreted-discharged filters were included in Oct 2004. The results were “disillusioning”, showing deviations between labs of 50-75%. Mr Schütz showed that the while the test stands could deliver reproducible results, the aerosol generators and the detectors used after the filters needed improvement. A future standard (ISO 21501) describes both the aerosol spectrometer detection method and the clean-room particle counter method. The latter is suspect because it is designed for use with very low particle levels, much lower than those obtained downstream of a cabin air filter.

Nanofibers for Defence

Seshastri Ramkumar of Texas Tech University has been applying nanofibers in the area of national defense and value-added products such as decontamination wipes, chemical protective clothing and highly efficient filters. Polyurethane and polyethylene oxide nanofibers in the range of 50 to 300 nm were electrospun and the first known self-assembling phenomenon was observed with the PU spun from 50/50 DMF/THF. These honeycomb-like nanofiber microstructures could be of value in chemical countermeasures substrates such as facemasks and chemical protective clothing liners. They appear when a nonconductive collector is used and seem to be the consequence of charged fibres repelling each other. Metal oxides embedded in polyethylene oxide nanofiber webs allow them to behave as self-detoxifying substrates due to their reactivity against chemical warfare agents, organophosphorous pesticides, and industrial toxic chemicals. PEO with Magnesium Oxide appears to break the P-O and P-F bonds in these chemicals. As an aside Mr Ramkumar commented that nanoparticles of gold are, unlike the bulk metal, reactive.

Simulating Nonwoven Filters

Andreas Wiegmann of the Fraunhofer Institute ( Germany ) described the development of a geometric model for the 3 dimensional structure of nonwoven. It contains the porosity, fiber diameters and fiber anisotropy as parameters and uses a random number generator to prevent no two realizations looking alike. In these structures, fluid flow and then particle deposition can be simulated. Initial pressure drop, filter efficiency and filter lifetime can be computed based on the structure, the mean flow velocity and the particle size distribution. Electric charges, adhesion forces and in case of very small particles, diffusion can also be modelled. With this simulation chain from nonwoven geometry to filtration properties, the virtual design of a nonwoven filter becomes feasible. Nonwoven deformations, strengths and acoustic absorption can also be simulated. Comparisons with measurements on real media show that the simulations are capable of reproducing real effects and have the potential to provide insights on new filter media design.

Simulating Permeable Nonwovens

Sabrina Zobel of NCRC has attempted to simulate a calendered spunbond using software that lays down a random array of rectangular fibres horizontally and vertically (90 degrees being the only angle of intersection present). In order to simulate bending on compression, each fibre is assumed to be made up of 14 micron cubes which shear by a predetermined percentage according to pressure applied. Validation against real spunbonds (2 denier fils) was carried out over a range of basis weights from 20 to 100 gsm. However all were calendared identically, so all had similar solid volume fractions. The simulation and the experimental results for permeability were indeed in the same region, but only over a narrow range of basis weights

Peptides as Biocides

Steve Arcidiacono of the US Army Natick R&D Center pointed out that t raditional biocides for protecting fabrics against degradation (e.g. metal and phenolic compounds) are neither eco-friendly nor selective toward the target organisms. Antimicrobial peptides, however, have biocidal activity against bacteria and fungi in microgram quantities and are safe for the environment, so Natick are checking them for biocidal activity in polymeric films and coatings. The antimicrobial peptide cecropin P1 has been shown to be active against bacteria and fungi in thin films and coatings even after exposure to organic solvents and high temperatures. These 20-40 angstom particles are obtained from toads, frogs, pigs and cows and when they come into contact with a bacteria they form an alpha helix, several of which self-assemble into a circle and burrow through the bacterial cell wall, puncturing it. The activity of treated films is diminished by washing, so Natick are now exploring grafting the peptides onto a surfactant molecule, one end of which will bond to the polymer, the other being free to form a helix and cooperate with its neighbors to kill bacteria. Asked about the likely shelf-life of peptides, this had yet to be checked, but informally, they've been in use for weeks without apparent loss of activity.

Hydrophilic Silane Quats

Tim Byers of AEGIS Environmental Management argued that the use of antimicrobials would soon be extended from their niche in socks, shoes and underwear into the more challenging fields of military uniforms, protective clothing and sportswear. Here reductions in odor were marketable, because odor leads to discomfort, and in the military case smelly soldiers can easily be detected at night. Conventional antimicrobials rely on leaching for their activity, but AEGIS have developed silane quats where the alkoxysilane bonds the quat to the fabric to give a “kill on contact” non-leaching activity. Unfortunately these compounds are normally hydrophobic and spoil the comfort of a garment, so AEGIS has developed a hydrophilic version for this application. The new version allows a treated cotton to behave like the untreated fabric in drop and wicking tests. On polyester it improves the wettability. Both cotton and polyesters with the new finish remain active for at least 25 washes. Asked about the durability of the original hydrophobic material, Mr Byers said it would do 1200 washes and was in reality only lost from the surface by abrasion.

Antimicrobial Nanofibers

Ilse Tuinman of TNO Defense Security and Safety Division ( Holland ) is developing antimicrobial filter layers by electrospinning inherently antimicrobial polymers. These filter layers will be incorporated into medical devices, military clothing and aerosol filters. The polymers used were modified quaternized acrylic polymers based on methacrylates and on polystyrene maleic acid anhydride. The methacrylates were prepared by copolymerization of dimethylaminoethylmethacrylate with butylmethacrylate in acetone at 85°C. After quaternization with alkylbromides C5 to C16 in acetone for 24 hours these polymers have shown a log 4 reduction and better against E.coli, Klebsiella pneumoniae and Staphylococcus aureus according to JIS 1902. The highest reduction obtained was log 5.5 for a polymer consisting of 84% butylmethylacrylate and an undisclosed compound developed by TNO (patent pending). PVA was also electrospun from water, and soaked in Dow 5700 antimicrobial. This showed almost no activity. Work continues with attempts to improve the uniformity and hence the filtration performance of the electrospun webs.

Copper Oxide Antimicrobials

Gadi Borkow of Cupron Inc. reminded us that copper and brass doorknobs kill E.coli where stainless steel has no effect. Copper water pipes kill Legionella bacteria and copper-bottomed ships go faster because of its natural anti-fouling properties. Copper ions damage cell walls and bacterial DNA without allowing tolerant strains to develop, so Cupron has developed copper oxide master batches to allow the production of fibers and films containing this safe biocide. Because the copper is a non-leaching kill-on-contact biocide, it has to be present at the polymer surface and this means in the skin of a bicomponent fibre. The resulting copper containing nonwoven fabrics possess potent antibacterial, antifungal and antiviral properties which are unaffected by washing. The metal has been used as a spermicidal contraceptive which can remain implanted for decades (as Inter-Uterine Devices) and fibers containing it are non-toxic, not irritant and non-allergenic. Impressive photos of diabetic ulcers cured by wearing Cupron socks were shown, and Mr Borkow went on to claim that Cupron face masks and pillow cases reduced skin wrinkles due to copper ions being stabilisers of skin proteins. Applications foreseen are:

• Disposable hospital textiles to reduce hospital acquired infections.
• Barrier fabrics and suits to protect the wearer and immediate environment from pathogens.
• Air filters to reduce the passage of viable microbes.
• Filters to reduce transmission of bacteria and viruses during transfusion of blood or blood related products.
• Diaper topsheet to reduce diaper rash infections.
• Masks to protect the wearer from aerosol transmitted pathogens, such as the avian flu.

Asked if the valency of copper was important, Mr Borkow said the actual additive was a proprietary mixture of Cu-1 and Cu-2.

Performance Evaluation

Ning Pan of the University of California at Davis has been using the PhabrOmeter he invented to get an index of the softness, smoothness and stiffness of a wide range of textile fabrics. The PhabrOmeter records the forces generated as a disc of fabric pinned down by a similar sized weight is pushed through a hole in the center the support disc. Fabric friction, stiffness, creaseability, stretch, stretch recovery and drape all affect the result and an analysis of the resulting force/time plot (“extraction curve”) can be used to quantify several aspects of fabric handle. If a fabric is put through the machine twice, the difference between the curves contains information on crease recovery. The curves obtained appear remarkably consistent for repeated testing of the same fabric. With regard to the thermal characteristics of the fabric – whether it is warm or cool to the touch being an important part of the subjective assessment - the density of the material was the key, and this affected the PhabrOmeter result. (Density or air content dominates the sensation of warmth c.f. the thermal conductivity or specific heat of the polymers used)

Evaluating Fabric Texture

Sachiko Sukigara of Niigata University ( Japan ) has correlated panel assessments of the handle of diaper topsheets, toilet papers and some electrospun nonwovens with results from the Kawabata method normally used to quantify the hand of conventional textiles. 40 samples of spunbond and 22 samples of double-ply virgin pulp toilet paper were assessed by 40 female judges using a 5-point scale. Then measures of tensile strength, shear, bending, friction, roughness and compression were obtained from the Kawabata system. Because finer fibers give more smoothness, two electospun nanofiber materials were also tested. These were 100nm silk fibers electrospun from fibroin in formic acid and PLA electrospun from DMF. Of all the measurements taken, one result, surface smoothness or to be precise the deviation of the coefficient of friction of the fabric surface, correlated best with the subjective assessment of handle of all the materials. For the electrospun materials the coefficient of friction was not as low as expected, maybe due to some stick-slip effects as the piano-wire probe was dragged over the surface. The deviation of the coefficient of friction was however very low. In response to questions, the electospun webs were 20 microns thick and lower than normal test pressure had to be used because they were easily damaged.

Absorbent Product Texture

Hiroko Yokura of Shiga University ( Japan ) has carried out a similar study, this time panel testing the surface of diapers and pantiliners both dry and wet and correlating the results with the Kawabata and surface dryness measurements. 68 different diapers were evaluated by mothers and students for softness, smoothness and wetness. Best correlation was between subjective assessment of wetness and the surface dryness results. For dry diapers and pantiliners, the deviation of the coefficient of friction was the best predictor of good handle. Ms Yokura thought this measurement could be used in quality control, where values below 0.045 could be regarded as acceptable.

PhabrOmeter Hand Assessment

Shulin Zhang of P&G's Fabric Care Division has also been comparing the PhabrOmeter with expert panel testing and the Kawabata system for fabric hand assessment. Tactile properties of fabrics and fibres (including hair) are important to several P&G operations so any method which can improve on panel testing or the complex Kawabata evaluation, or give similar data more quickly, would be valuable. Terry towelling treated with various fabric softeners provided a range of textures to be assessed by expert panel and Kawabata. The fabrics were then evaluated on the PhabrOmeter and the “extraction curves” analysed to find the best correlations. Correlations between the Kawabata coefficient of friction and both the maximum curve height and the area under the curve (“extraction energy”) were excellent, but the curve height correlated less well with the panel assessment of softness. Here the initial slope of the curve and the extraction energy corellated best. If “Extraction Energy Reduction” was calculated by subtracting the EE result of a treated fabric from the EE result of the untreated (unsoftened) control, this EER correlated best of all. Dr Zhang concluded that the PhabrOmeter provided a valuable additional tool for tactile assessment, but should not be thought of as a replacement for panel testing or Kawabata. In response to questions:

• Wet fabrics could be tested on the PhabrOmeter but stiffness and friction may lead to high forces.
• The extraction curve has a maximum because the more the fabric is pushed into the hole, less is gripped by the weight.
• It works well with stretchy materials (Most materials are stretched in extraction)

Shouldn't hand be assessed against a soft material rather than smooth metal? Dr Zhang works mainly with cotton so these effects are not important. Anyway, the PhabrOmeter is not a replacement for hand testing.

Hydroentangled Delivery Systems

Steve Russell of the University of Leeds (UK) recycled the paper on 3D Nonwovens given at Pira's Future of Wipes Conference (July 2006)

Future of Spunlaid

Hans Geus, Technical Director of Reifenhäuser GmbH ( Germany ) went through the usual introductions to the technology and market, the statistics for which showed that spunlaid had lost share to drylaid between 2003 and 2004 in Europe . He reminded us that if we thought calendered spunlaid fabrics were flat and thin we should remember we could increase their thickness by several means:

• Through-air bonding instead of calendering
• Hydroentangling either after or instead of calendering
• Mechanincal softening

So, if the market requires the bulk and softness of dry laid, the future of spunlaid will be in different bonding systems. Furthermore if squarer fabrics are required for technical applications, these can be made by altering the aerodynamics of the laydown technique. In response to a question, spunlaid output was still growing at 9% per year, 5% of this being in the developed world.

Nanofibers from Spunbond

Nataliya Fedorova of NCSU has spun islands-in-a-sea bicomponent fibers with 360 Nylon islands in a PLA sea. If the ratio of nylon to PLA is 25/75, and the PLA is dissolved in caustic soda, nylon filaments with a diameter of 360 nm are liberated. At a more reasonable 75/25 nylon/PLA ratio they are half a micron in diameter. Tenacity (5 gf/denier) and modulus (140 gf/den) of the finest fibers is surprisingly good compared with the 100% nylon 16 micron homofil obtained from the same system. As for regular fibers, the strengths increased as the filaments got finer. Fabric properties after hydroentangling followed by PLA removal showed the nanofibre nonwovens to have substantially higher tear strengths (x3) than the 16 micron homofil nylon control. Tensiles were however only about half that of the control. Calendering the hydroentangled webs before PLA removal resulted in large strength losses, presumably because the PLA interfered with the nylon/nylon bonding. (Fabrics were tested after PLA removal). As Ms Fedorova concluded, this does indeed look like a promising technique for making tough microfibre fabrics for industrial applications.

Nanofibers as Sensors

Kris Senecal of the US Army Natick Soldier Center described the generation high surface area nanofibrous membranes with covalently attached Molecular Recognition Elements (e.g. antibodies, peptides/DNA) for the selective binding/capture of biological agents or food pathogens. The association of MREs with high surface area electrospun fiber membranes could allow the development of sampling devices tailored to specific hazards. Polymers with carboxylate or amine functionality (PVC-COOH or Polyurethane) were electrospun onto a stainless steel mesh, and the MRE's attached. For example, Avidin was chemically crosslinked to the membrane and coupled with biotinylated anti-staph antibodies. Membranes thus treated were shown to capture a 1 ng/ml concentration of Staph Enterotoxin B, despite the use of a non-optimised system. For the future, Natick intend to develop nanofibers not only for capture, but also to allow a capture event to generate an electronic signal through the membrane. To this end conductive versions of the sensors mentioned above have been created by polypyrrole coating of the nanofibre web.

Temperature effects in Meltblowing

Randall Bresee of the University of Tennessee recalled the contradictory reports on the effects of temperature on fiber diameter in melt blowing and has therefore conducted a series of experiments on a 15 cm lab line and a 115 cm commercial line. He has now shown:

• Increasing process temperature reduced the mean, minimum and maximum fiber diameters in webs but increased fiber diameter variability.
• The influence of temperature on fiber diameter was less at higher process temperatures than at lower process temperatures.
• Fiber contact/fusion depended on process temperature.
• Contact/fusion was thought to contribute to the increased diameter variability that was observed when process temperature was increased.
• Primary air temperature was more important than die block temperature. (On the commercial line the die block was heated from 180 to 280C without affecting fibre diameter – air temperature being constant at 200C)
• Die tip temperature was more important than die block temperature.
• Sensitivity to temperature change reduces as throughput increases. (more hot polymer damps out air temperature effects)

In response to questions:

• There was no shot formed at low temperatures.
• There was no fly (fine fibres being lost) even at 280C.
• There was no degradation of the PP even at 280C (residence time too low).
• MFR's used were 1100 to 1200. At higher levels degradation would probably occur at the higher temperatures.

Improving Filtration Performance

Mark Snider of Nordson Corporation described a new melt blowing process intended (eventually) to allow the production of nanofibres. In essence this is a bico-meltblown process producing segmented fibers which separate as a result of careful choice of polymers and optimal setting of temperature and air flows. At present the average fiber diameter (~0.75 micron) appears to be about half that achievable with conventional melt blown so maybe we are looking at side-by-side bicomponency. If the line is set to produce 1-2 micron average diameter, throughput can be double that of conventional melt-blown. Measurements of hydrohead on a range from 10-20 gsm fabrics showed that the new system delivered double that of the old. For the future, Nordson will be trying to get the average diameter down to 0.5 microns and will be optimizing polymer additives and electreting to allow the production of high efficiency filters. Asked about the throughput from the new system, Mr Snider said that 0.17 to 2 gms/hole/min were now possible compared with 0.1 g/hole/min on the old system (for 1-2 micron fibers)

Hydroentanglement optimization

Prof. Memis Acar from Loughborough University (UK) reviewed the last few years of work aimed at optimizing the nozzle geometry, process conditions and energy efficiency of the hydroentanglement process. His conclusions were as follows:

• Cone-up nozzles with a wide range of cone angles are sensitive to water pressure and give lower jet velocities but high discharge coefficients compared with cone-down nozzles. Cone-down nozzles give the longest intact water column length and are more efficient.
• Parallel laid webs are stronger in the machine direction.
• Tensile strength increases with pressure up to a maximum dependent on web density and fibre type (the critical pressure) and then declines with further pressure increases.
• At or above the critical pressure, process speed has little effect on the strength. Below the critical pressure, higher speeds give weaker fabrics especially for high modulus (hard to entangle) fibers.
• Fabric strengths increase sharply when the web is entangled from the opposite side.
• Prolonged entanglement from one side is less efficient than briefer entanglement from alternate sides.
• Choice of pressure profile is crucial for efficient entanglement, and this will vary with fibre type and web structure.

Image analysis of the orientation and distribution of straight fibre segments in entangled webs correlates with mechanical properties .

CFD simulation of HE nozzles

Nagendra Anantharamaiah of NCRC reminded us that hydroentanglement efficiency depends on the waterjets maintaining their intact length for an appreciable distance downstream of the nozzle. This is achieved by using a cone-down nozzle where the sharp 90 degree inlet forces the water to detach from the nozzle wall (the “hydraulic jump”) and remain unaffected by cavitation or wall-induced turbulence. As the sharpness of the inlet is eroded by the high pressure water, efficiency diminishes and eventually the hydraulic jump collapses allowing higher discharge coefficients but lower intact lengths. In practise, visible (under the microscope) rounding of the inlet is evident after 48 hours running at 3000psi. In his current work, Dr Anantharamaiah has simulated the steady-state in the nozzles using CFD and has investigated the effects of increasing inlet roundness on the nozzle's discharge coefficient. Discharge coefficients obtained from these simulations were indeed found to increase with increasing the inlet roundness and agreed with practical data obtained from measuring the inlet roundness with 3D profilometry over the range 0 to 0.2 r/d. A simple equation for predicting the nozzle inlet roundness by measuring its discharge coefficient at high pressures was therefore obtained.

Removing jet streaks

Katharina Rompert of Freudenberg ( Germany ) showed how adding a second row of holes to a hydroentanglement nozzle strip could eliminate the jet streaks commonly seen on hydroentangled fabrics. This second row could use smaller holes, and they had to be slightly offset so that the water columns impacted the fiber “peaks” raised at the sides of the first row's jet streaks. The improved surface uniformity was clearly visible in the photographs shown and Ms Rompert had also carried out a thorough image analysis to prove it and quantify it. The effect of the second row of holes was most pronounced when entangling on a belt at 200 bar but on a drum some stripiness was still visible. Various 2 nd row hole sizes were tried, 110 microns (following a 130 micron 1 st row) being best. Moving to 4 rows of holes from two gave no further benefit. With 2 rows, improved tear strengths were obtained but tensiles were unaffected.

3D Modelling of HE

Ping Xiang of NCSU presented a model of the hydroentanglement process based on the first principles of fluid mechanics. The CFD domain used in the simulation was a rectangular space with a width of the distance between the centres of two adjacent jet holes and a length of 0.45 mms in the MD. Three layers were assumed in the vertical direction, the first a layer of water of depth equal to twice the fibre diameter, the second being the fibre web (0.7mm) and the third being the forming wire. The forming wire is assumed to move at 2 m/s. Impact between the turbulent jets and these layers creates further turbulence and the fibres are entangled in the resulting vorticity. Tensile strengths obtained from fabrics entangled at different pressures were plotted against the vorticity calculated by the CFD package at these pressures. The correlations were near perfect, and said to validate the initial assumptions that tensiles would be proportional to degree of entanglement, and that entanglement was caused by fibres moving around in the vorticity. Furthermore, simulations of the effects of different hole sizes gave 0.127mm as the optimum size for maximum vorticity, this size being said to be pretty close to that adopted in industrial practise.

Dyeing of HE nonwovens

Andrew Hewitt of Leeds University (UK) recycled the paper he gave at EDANA's Nonwovens Research Academy in April 2006. The same concerns related to the lightly entangled nonwovens ability to survive commercial dyeing conditions were raised in questions.

Calvin Woodings - 3 rd October 2006

No comments: