The current need in modern R&D facilities to produce more results per unit of calendar time at equal or lower cost has boosted interest in high-throughput experimentation technologies. It is highly desirable to have machines (robots, workstations, and the like) to carry out elemental manipulations that are now carried out by laboratory staff. When one is able to control the order and nature of these elemental manipulations, the unique set of experimental steps carried out in non-routine R&D laboratories can be automated. This work will then be possible in less time (in theory such machines continue to produce results for 24 h per day and seven days per week) and often the outcome is a better reproducibility than when human staff is carrying out these experimental procedures.

Automated manipulations are difficult, if not impossible, when highly viscous liquids are used. In handling viscous liquids to make series of samples, the problem is to find an automatic dispensing system (or auto-pipetting system) that is able to handle these high-viscous media with good reproducibility and within a short period of time. Reproducibility is a major problem particularly when handling viscous liquids, and even complete clogging of the system may occur, impairing a practical application of automated systems for high-viscosity liquids in high-throughput experimentation and the like.

By"high-viscous"are meant solutions, liquids, suspensions, and gelled materials with a viscosity during dispensing of 250 mPa. s or more, preferably measured at the temperature at which the material is dispensed.

In combinatorial experimentation, often a positive displacement syringe pump is employed. For example, US 2002/0159919 describes a method for high-speed precision dispensing of microfluidic quantities of reagents and other liquids with a wide dynamic range of dispense volumes using this type of pump. WO 00/66269 discloses a method for sampling fluids in amounts of less than 5 microliters per aliquot to a dispense element array using a positive displacement syringe pump such as a CAVRO syringe pump. US 2001/0016631 also mentions the use of a syringe pump or gear pump in combinatorial experimentation. However, this type of pump is not suitable for handling high-viscosity liquids.

In handling complex liquid systems such as concentrated surfactant formulations or polymer-containing liquids, all kind of rheological behaviour is encountered, and regularly visco-elastic systems, thixotropic systems, etc. are extremely difficult for standard automated pipetting equipment known in the art to handle automatically.

In the currently known auto-pipetting systems those complex types of liquids often cannot be handled.

Various conventional pumps are known to be suitable for pumping viscous material. US 2002/0100770, for example, describes positive displacement dispensing apparatus for dispensing precise quantities of a viscous liquid of less than 1 mm3 in size. This liquid may be an adhesive with a viscosity of 10,000 to 1,000, 000 centipoise. Usually, gear-wheeled pumps are used to pump viscous materials. However, when used for making an array of samples comprising the step of automatically dispensing a liquid with a viscosity under dispensing conditions of at least 250 mPa. s for automated high-throughput experimentation, gear-wheeled pumps suffer severely from wear and are insufficiently resistant to chemicals. Moreover, their accuracy is insufficient. It was also found that the valves of gear-wheeled pumps get stuck in viscous masses having a viscosity higher than 500 mPa. s. Also plunger pumps have similar problems regarding wear, chemical

resistance, and valve problems. Membrane pumps, which are also regularly used for pumping viscous materials, also suffer from valve problems, as do hybrid membrane-plunger pumps. In addition, peristaltic pumps were found to be suitable to dispense viscous liquids, but their accuracy and therefore their reproducibility is very poor, rendering them unsuitable for application in automated systems.

In the few cases wherein high-viscous liquids are handled, the volume of liquid dispensed is significantly lower than the intended volume, and at least cannot reproducibly be dispensed in the vials without contamination of e. g. system liquids.

The differences in dispensed volumes depend on the rheological nature of complex liquids and the amount of time used for aspiration and dispensing, which make these conventional systems far less suited to automatic dispensing of viscous liquids as for example required for high-throughput experimentation. Moreover, due to the above-mentioned problems, samples cannot be dispensed within the short period of time required in efficient automated high-throughput experimentation.

The skilled person generally knows the difficulty of automatic handling and administering of viscous materials, see for instance B. Chrisholm et al., Prog. Org.

Coat. 2002,45, 313-321.

Therefore, there is a need for an automatic dispensing system (or auto-pipetting system) that is able to handle high-viscous media with good reproducibility and accuracy within a short period of time.

It has now surprisingly been found that a method of making an array of samples by automatically dispensing high-viscous media with good reproducibility and accuracy is available when using a positive displacement rotary piston pump.

Thus, according to the present invention, there is provided a method of making an array of samples comprising the step of automatically dispensing a liquid with a viscosity under dispensing conditions of at least 250 mPa. s, preferably at least 500 mPa. s, more preferably at least 750 mPa. s, by using a positive displacement rotary

piston pump. Said viscosity is preferably measured in the conventional way at the temperature at which the material is dispensed. If said temperature is unknown, it may be preferable to measure at a temperature of 23°C.

The method of the invention is particularly useful in high-throughput experimentation using arrays of samples or reaction vials. Such an array preferably comprises at least three vials, and usually substantially more than three. The method is suitable for high-viscosity liquids, i. e. having a viscosity of 250 mPa. s or more, even up to 100 Pa. s. Liquids with a higher viscosity can be applied, however, this is done mostly at the cost of accuracy and reproducibility. The method of the invention further is suitable for fast dispensing of the high-viscous liquid, and each sample in the array can easily be dispensed at a dispense rate of at least 500 pI/min, using a pipetting robot system and a robot gripper arm, so that this experimental set-up can prepare arrays of samples. Preferably, the dispense rate is at least 1,000 pI/min, more preferably at least 2,500 pI/min, and most preferably at least 5, 000 ul/min. These arrays of samples can be used in combinatorial experiments. This set-up is automated so that simple programming of the experimentation steps can be carried out, allowing an automated procedure to prepare these arrays of samples.

In a preferred embodiment according to the invention, the method comprises a step of aspirating the viscous liquid using tubing with a large diameter, at least 5 mm in meter, and preferably dispensing it using a conical dispense tip. This embodiment further adds to the accuracy and reproducibility of the method.

According to the present method, the automated arm either brings the dispense tip to the array of sample vials, or it moves one or more sample vials to the dispense tip. It further is an advantage for obtaining accurate dosing when the dosing amount is determined gravimetrically. In that case, the vials are placed on a

balance, and the pump speed is controlled by a protocol using the target weight and the actual weight as parameters.

The method of the invention can be operated due to the selection of a suitable pump, i. e. a positive displacement rotary piston pump. Such pumps appear to be suitable for accurate dispensing of high-viscous liquids in high-throughput experimentation, which dispensing up to now was not possible. In principle, any positive displacement rotary piston pump can be used, as long as its capacity is sufficient for the purpose for which it is used. Preferred positive displacement rotary piston pumps are rotary-axial piston pumps, such as can be obtained at LaboCat (The Netherlands) under the trade names HPLH PCON-CT"", CAT micrometering pump, and the like. The pump may be used in combination with software for flexible control, such as CAT LabControlTM software. The parts of the pump are made of materials that are resistant and inert under the conditions used, such as ceramics, PTFE (polytetrafluoroethylene), and stainless steel.

In order to allow for quick and time-efficient cleaning of the set-up, all tubing and dispensing tips are mounted in an easy to remove manner. To obtain reproducible dispensing volumes, the tubes used for loading the pump system (aspiration tubes) and the tubing for connecting the pump and the dispense tip preferably have an inner diameter larger than 3 mm, more preferably larger than 4 mm, and most preferably larger than 5 mm. Thinner tubing results in a significant loss of reproducibility and an increased dosing time.

Furthermore, in order to achieve a better reproducible dispensing volume, the material used for the dispense tubing should preferably be relatively rigid, so the internal pressure in the tubing cannot be compensated for by a volume increase caused by the elasticity of the tubing. Preferred material for making the tube is PTFE such as Teflon@, which is preferred over silicone rubber tubing.

Finally, it is noted that the throughput can be increased by using a multichannel instead of a single-channel pump. In this manner, it becomes possible to dispense

more than one component either sequentially or simultaneously in a single vial without pump cleaning.

The invention is further illustrated by the following unlimitative examples.

Example 1 An experimental set-up was used consisting of a rotary piston pump (CAT", pump type: HPLH 300 VCS).

Viscous liquid : Setamine US-138 BB-70 (ex Akzo Nobel Resins) Viscosity: 1.2 Pa. s (23°C, 100 s~) Dispensing speed: 25 ml/min Amount : 10 mi I. Run 1: Tubing used: about 10 cm Teflon aspirating tube with 6 mm diameter and 30 cm silicone dispensing tube with 6 mm diameter.

II. Run 2: Same as run 1, but a dosing tip with 1 mm diameter was mounted at the end of the dispensing tube.

Output of array of samples in grams (density 1.0 g/l) Table I : I. (grams) I1. (grams) 9. 98 9. 98 9. 97 9. 98 9. 96 9. 98 9. 96 10. 00 9. 98 10. 00 This example shows the accuracy of the system.

Example 2 This example was performed as Example 1, with the following constituents.

Viscous liquid : Resin type Setalu) 6100-GR-74 (ex Akzo Nobel Resins) Viscosity: 20 Pa. s (23°C, 100 s~1) Dispensing speed: 2 ml/min Amount: 1. 0 ml In this example, a silicone dispensing tube with 3 mm diameter was employed.

I. Run 1: Tubing used: about 10 cm Teflon aspirating tube with 6 mm diameter.

II. Run 2: Same as run 1, but the dispensing speed was lowered to 1ml/min, and a tip with a diameter <1 mm was mounted on the dispensing tube.

Output of array of samples in grams (density 1.0 9/l) Table li : 1. (grams) li. (grams) 0. 88 0. 97 0. 96 0. 96 0. 96 0. 99 0. 91 0. 99 0. 89 0. 97 This example shows that when tubing with a smaller diameter is used, the required accuracy and reproducibility could be achieved by lowering the dispensing speed and dispensing the viscous liquid into a conical dispense tip.

Example 3 This example was performed as Example 1, except in combination with a Lipos (Zinsser Analytic) automated liquid and powder dispensing system, which consists of a robot gripper for picking up vials and trays of vials to make a series of samples.

Viscous liquid: Resin type: Setals 90173-SS-50 (ex Akzo Nobel Resins) Viscosity: 20 Pa. s (23°C, 1 s~1) or 0.80 Pa. s (23°C, 100 s~1) In both runs the same tubing was used: 10 cm Teflon aspirating tube with 6 mm diameter and 30 cm Teflon dispensing tube with 6 mm diameter. A tip with a 1 mm diameter was mounted onto the dispensing tube.

Dispensing rate 10 ml/min, volume 10 mi II Dispensing rate 20 ml/min, volume 20 ml Output of array of samples in grams (density 1.0 g/l) Table III : Sample { 1 9. 71 19.48 2 9. 70 19.51 3 9. 73 19.48 4 9. 76 19.49 5 9. 76 19.50 From Example 3 it can be seen that when tubing with a large diameter is used, and the viscous liquid is dispensed into a conical dispense tip, excellent accuracy and reproducibility are achieved even at high dispensing rates.

Example 4 This example was performed as Example 3. In this experiment the relation between dispensing accuracy and set volume was investigated.

For set volumes up to 1,500 pLI a dispensing rate of 2,000 pI/min was used. For higher set volumes a dispensing rate of 8,000 pI/min was used.

Viscous liquid : resin type Setalux 1200-XX-55 (ex Akzo Nobel Resins) Viscosity: 1.6 Pa. s (100 s~, 23°C).

Table IV : Set volume Actual % (ml) weight (g) 0.06 0.05 83 0.52 0.50 96 0.55 0.52 95 0.58 0.55 95 1.28 1.23 96 1.30 1.25 96 2.60 2.48 95 2.62 2.50 95 3.97 3.75 94 4.00 3.78 95 4.50 4.26 95 4.75 4.50 95 4.80 4.56 95

In Table IV it can be seen that, independent of the set volume, high accuracy values are obtained.

Example 5 This example was performed as Example 3, dosing a viscous gel.

Viscous gel : 2 wt. % xanthane.

From Table V it can be derived that even when dosing a viscous gel, accurate and reproducible results are obtained.

Output of array of samples in grams: Table V: Speed 10 mi/min 1 ml/min Set volume 2. 00 ml 0.50 ml 1. 93 0. 48 1. 90 0. 49 1. 92 0. 47 1. 89 0. 49 1. 94 0. 47 1. 95 0. 50 1. 94 0. 48 1. 93 0. 48 1. 94 0. 47 1. 93 0. 48 Comparative Example 6 An experimental set-up was used comprising a 2.5 ml syringe pump with ethanol as system liquid. The sample was aspirated and dispensed through a standard 158 mm pipetting needle.

Viscous liquid : glycerol 100% Viscosity: 1.76 Pa. s (23°C, 100 s'')

Density: 1.26 g/ml It was found that when a syringe pump is employed, the dispense volume was appreciably less than the set volume. Moreover, contamination with system liquid was severe: 7-23 % by weight (see Figure 1).

Example 7 In this experiment a series of viscous liquids with different viscosity ranges (all measured at 23°C, 100 s~1) were dispersed. In a first experiment, liquids with a viscosity of 1 Pa. s, a viscosity in the range of 5-10 Pa. s, and a viscosity in the range of 10-20 Pa. s, respectively, were dispersed in volumes of 10 ml per aliquot.

The maximum flow rate was varied from 25 to 1 ml per min. The relative standard deviation (rsd) in dispensed volume (in %) was determined after 5 measurements indicating the deviation in relation to the, on average, dispersed volume. In a second experiment, liquids with a viscosity in the range of 5-10 Pa. s were dispersed using a maximum flow rate of 5 ml/min and the dispense volume was varied from less than 0.3 ml to more than 1. Again the rsd was determined. The results of both experiments are summarised in Table V). As can be clearly seen in said table, accuracy and maximum flow rate depend on the dispense volume and the viscosity of the liquid or gel to be dispersed.

Table VI Viscosity Volume Max flow rate, Rsd (Pa. s) ml mllmin (%) 1 10 25 <0. 2 5-10 10 5 < 1 10-20 10 1 <2 5-10 <0. 3 5 10 5-10 0. 3-1 5 < 2 5-10 >1 5 <1