Progress in Organic Coatings 38 (2000) 193–197

A rheological characterisation technique for fast UV-curable systems

Sang Sun Lee, André Luciani, Jan-Anders E. Månson∗

Département des Matériaux, Laboratoire de Technologie des Composites et Polymères (LTC), MX-G Ecublens,Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland

Received 14 September 1999; accepted 16 March 2000

Abstract

Using a rheometer coupled with a UV-light generator, a photo-rheometry set-up has been developed to study the viscoelastic propertiesof UV-coating systems during fast curing. Due to a high reaction rate, the viscoelastic properties have to be evaluated using a specialprocedure. This technique was found suitable to obtain reliable rheological data during the fast photo-reaction, allowing the determinationof gel points occurring within less than 1 s. © 2000 Elsevier Science S.A. All rights reserved.

Keywords:Rheology; UV; Gel point; Coatings

1. Introduction

UV-curable systems are now widely used as coatingsmainly because of their low curing temperature, the absenceof organic solvent and low dimensional shrinkage. The in-creasing efforts in the UV-curable powder coating industryis devoted to the development of new formulations or pro-cesses to provide superior mechanical properties and surfacecharacteristics together with a high curing rate. Today, theformulation of UV-curable systems is usually done on anempirical basis. The aim of the present study is to contributeto the development of characterisation techniques adaptedto the particular behaviour of these systems.

Because the coating process as well as the ultimate filmproperties are strongly affected by the rheological behaviour[1,2], the evolution of the viscoelastic properties during cur-ing are important characteristics of a UV-curable powderformulation. One of the distinct characteristics of industrialUV-curable systems is the high reaction rate taking placeunder intense UV radiation, the entire reaction usually tak-ing place in less than a few seconds [3]. Systems showing50% conversion within less than 0.3 s are now commonlyused [4].

Such fast reactions are very difficult to follow and anal-yse using classical methods and equipment. Most of theanalytical methods used to monitor the cure profiles canbe classified into two main categories. The first categoryis based on a time consuming process of measuring the

∗ Corresponding author. Tel.:+41-21-693-2951; fax:+41-21-693-5880.E-mail address:jan-anders.manson@epfl.ch (J.-A.E. Månson)

degree of cure before and after short UV pulses. Infraredspectroscopy [5] and nuclear magnetic resonance [6] havebeen used in this context but can lead to erroneous interpre-tation because of the dark reaction which may occur afterUV exposure. The second category is based on continuousmonitoring of the curing reaction with techniques such asdifferential photo-calorimetry [7], dilatometry [8], dielec-tric relaxation spectroscopy [9] or real time ATR–FTIR[10], which give very useful information about the reactionkinetics. Viscoelastic analyses have also been applied tofollow curing kinetics but only for systems with relativelyslow curing rate [11].

The purpose of this work is to develop a reliable methodto measure the viscoelastic properties of fast photo-reactionsystems during curing. This was done by combining a com-mercially available rheometer with a UV-light generatorto measure the rheological properties during cure and inparticular to determine the gel point.

2. Experimental

2.1. Materials

A UV-curable clear coat powder sample from HerbertsPuverlacke GmbH (Germany) was selected to test the poten-tial of this method. According to the manufacturer, the mate-rial is made of an unsaturated polyester acrylate resin with aTg of 40◦C and contains a small amount of flowing additiveused to improve the levelling of the material after melting.The photo-initiator used in this work was dimethoxy phenyl-

0300-9440/00/$ – see front matter © 2000 Elsevier Science S.A. All rights reserved.PII: S0300-9440(00)00088-6

194 S.S. Lee et al. / Progress in Organic Coatings 38 (2000) 193–197

Fig. 1. Complex viscosity and loss modulus as a function of the frequencyat 110 and 120◦C.

acetophenon (Irgacure 651, Ciba Speciality Chemical) at aconcentration of 1 wt.%. Titanium dioxide is usually incor-porated in the formulation of this commercial material, butwas not used in the present study because the presence ofa filler changes the UV absorption, and makes it difficultto interpret the rheological data. Before testing, the mate-rial was pressed into plate samples at 110◦C to avoid thepresence of air bubbles. The rheological characterisation ofthe initial material was done on a controlled strain dynamicrotational rheometer, Rheometrics RDA II, equipped with50 mm parallel plates with a 1 mm gap between plates. Re-sults obtained at 110 and 120◦C are depicted in Fig. 1, show-ing a nearly Newtonian behaviour over the frequency rangeinvestigated. The measurements were checked to be in thelinear viscoelastic range by doing strain sweep experimentsprior to the dynamic frequency sweep experiments.

2.2. Photo-rheometry equipment

The rheological analyses during curing of the sampleswere conducted in the same rheometer combined with aprecision Novacure UV-light generator. Fig. 2 shows aschematic view of this photo-rheometry set-up. The top andbottom fixtures are connected to the transducer and actuatorof the rheometer, respectively. The upper fixture incorpo-rates a removable quartz cylinder allowing irradiation of thesample without modification of the spectral characteristicsof the original UV light. The UV light was transmitted tothe rheometer set-up using a liquid light guide. The materialto be tested is placed between the two 8 mm parallel platesand UV light is directed onto the sample by a mirror placedat a 45◦ angle to the upper fixture. In our experiments, thesteel bottom plate was cleaned after each experiment tomaintain the reflectivity constant. All the results presentedin this work were carried out in isothermal conditions, ei-ther at 110 or 120◦C. The sample thickness was chosen tobe constant, 150mm, and the diameter of the plates was

Fig. 2. Schematic diagram of the photo-rheometry set-up.

always 8 mm except for the initial characterisation shownin Fig. 1. The UV intensity was measured to be about15 mW/cm2 (between 320 and 390 nm wavelength) at thesample surface.

2.3. Rheological characterisation of the curing reaction

Apart from the evolution of the classical rheological func-tions during cure, the gelation time is probably the mostimportant characteristic of a curing system, since it definesthe transition between the liquid and the solid state of thematerial. From a practical point of view, this transition is ofprimary importance because no more flow can be inducedin the sample after the gel point without breaking the mate-rial, at least partially. This means that all the processing andforming steps of an industrial process should be completedbefore it occurs.

Several rheological definitions have been suggested todetermine the instant of gelation [12–14]. In fact the onlymethod which is substantiated by reasonable theoretical ar-guments is that proposed by Winter and Chambon [15,16].According to those authors, the gel point can be determinedfrom dynamic measurements by observing the frequency de-pendence of the loss tangent, the gel point being determinedwhen this parameter becomes independent of the frequency.Another technique, which is very often employed, locatesthe gel point at the crossover between the storage and lossmodulus [17,18]. Except for a specific case, this method hasbeen shown to lead to erroneous results by very simple argu-ments like the frequency dependence of the gel point loca-tion which is, by definition, a chemically induced structuralchange independent of the frequency.

Despite this theoretical and practical inapplicability, thistechnique is nevertheless often used to characterise curablesystems. Another rheological technique has been proposedwhere the gel point is determined from an increase of thestorage modulus above an arbitrary value [19]. Due to thisarbitrary choice and absence of real theoretical justification,

S.S. Lee et al. / Progress in Organic Coatings 38 (2000) 193–197 195

even if this technique can give a mean of qualitative com-parison, it cannot be applied to a quantitative determinationof the gelation.

3. Results and discussion

Fig. 3 shows the evolution of the complex viscosity ofthe UV system used in this study as a function of irradiationtime. The measurement was done at a frequency of 62.8 rad/s(10 Hz) at 120◦C with a deformation amplitude of 3%, theUV exposure beginning at timet=0.

Usually, the procedure used to extract the viscoelasticfunctions from the strain–stress response involves compu-tation on several periods of measurement with the assump-tion that the system is stable during measurement. With ourequipment and material, one can see (Fig. 3) that the sam-pling rate was typically limited to about 1 s, which is obvi-ously not short enough to obtain information on such fastcuring systems, where most of the reaction has already oc-curred during this time interval. If this acquisition limitationcan be acceptable for a global analysis of the reaction, itprevents any attempt at gelation analysis, since it happensgenerally in the early stages of the reaction.

In order to get information about gelation, Khan [20] sug-gested a method where radiation pulses are applied to thesample with a frequency sweep test performed between twosuccessive pulses. However, in the present study, this methodwas not applicable because the system was shown to evolvedifferently with different UV irradiation history, especiallywith short UV pulses. Fig. 4 shows the evolution of the com-plex viscosity,η* as a function of time and number of pulsesfor different pulse lengths and Fig. 5 depicts the increase ofviscosity as a function of the cumulative irradiation time.Above 1 s, the pulse duration does not seem to influence thereaction as shown by the viscosity curves and gives similar

Fig. 3. Evolution of the complex viscosity with time under UV irradiation.The test was done at 120◦C with a frequency of 62.8 rad/s and a strainamplitude of 3%.

Fig. 4. Evolution of the complex viscosity with UV-pulse irradiation at120◦C.

results as for continuous irradiation. However, in fast curingsystems it is important to get information in relatively shorttime (∼0.1 s) to obtain a good estimation of the gel point,which means that a pulse technique is not suitable to oursystems.

To circumvent this problem, the analysis was done bydividing the strain and stress signals into very short period oftime (0.02–0.05 s) and by fitting sinusoidal functions to bothdata sets, assuming that the system is stable during this shorttime period. In a typical experiment an acquisition frequencyof 200 measurements per second was used. The viscoelasticfunctions are then calculated by a separated computationafter the input and output sinusoidal waveforms have beendirectly recorded from the photo-rheometer.

Fig. 6 shows the evolution of the stress response during afew seconds after irradiation at 62.8 rad/s with a 2% strainamplitude and a temperature of 110◦C. Again, for all thefrequencies the strain amplitude was checked to be small

Fig. 5. Evolution of the complex viscosity with cumulative time ofUV-pulse irradiation at 120◦C.

196 S.S. Lee et al. / Progress in Organic Coatings 38 (2000) 193–197

Fig. 6. Strain–stress response of the UV-curable system in the initial stageof irradiation.

enough to ensure that the measurements were done in thelinear vicoelastic range during all the reactions.

Fig. 7 presents the loss and storage modulus computedfrom the stress–strain data at a frequency of 62.8 rad/s at 110and 120◦C. It should be noted that due to the low thicknessof the sample between the parallel plates, an important errorcan be made for this dimension. The recommended thicknessgiven by the manufacturer is between 0.5 and 2 mm. In ourexperiment, an imprecision of 15mm leads to 10% error inthe thickness estimation and, without a perfect alignment ofthe parallel plates, this error can exceed this value.

The effect of this error can be seen on the initial values ofthe viscoelastic functions obtained in Fig. 7 in comparisonwith the results obtained in Fig. 1 with measurements doneon the 50 mm parallel plates with 1 mm gap between plates.The difference observed between theG′′ values are in theorder of 20%.

Since the goal of this study was not to study in detail theviscoelastic behaviour of the material during cure but ratherto show the applicability of the technique, further attempts

Fig. 7. Storage and loss modulus,G′ and G′ ′, of the UV-curable systemas a function of time under UV irradiation at 110 and 120◦C.

Fig. 8. Evolution of tanδ as a function of time for three different fre-quencies at 110◦C. The crossover point corresponds to the gel point.

to reduce the error were not done but a higher precision inthe viscoelastic functions could undoubtedly be obtainedwith a more careful set-up of the test geometry. To bestmimic industrial conditions, the thickness of the sampleshould be below 150mm to obtain an homogenous curing,which means that a very accurate set-up is necessary to ac-quire information about the viscoelastic functions with verythin sample. Even though this limitation is only of technicalnature, it can be nevertheless an important drawback forthickness of the order of 10mm with a standard commercialrheometer.

If the viscoelastic functions are directly dependant onthis experimental error, the gel point determination is sup-posed to be independent of it since it is obtained from thetanδ curves for which the error is eliminated by theG′′/G′ratio. The evolution of tanδ with time at three differentfrequencies at 110 and 120◦C are shown in Figs. 8 and9. The crossover points are located at about 0.9 and 2 s at120 and 110◦C, respectively, showing how an increase intemperature decreases the gel time.

Fig. 9. Evolution of tanδ as a function of time for three different fre-quencies at 110◦C. The crossover point corresponds to the gel point.

S.S. Lee et al. / Progress in Organic Coatings 38 (2000) 193–197 197

4. Conclusion

Using a commercial rheometer coupled with a UV-lightgenerator, a rheological technique was developed to mea-sure the viscoelastic behaviour of UV-curable systems dur-ing cure. This technique is well adapted to the evaluation ofthe rheological properties of ultra-fast photo-reaction sys-tems and particularly to the determination of the gel point.In a future work, the influence of the physical parameters(UV intensity, pulse radiation, film thickness, temperature,deformation amplitude, etc.) as well as the chemical charac-teristics of the materials on the cure profiles will be studiedusing this technique.

Acknowledgements

This research was supported by Herberts PuverlackeGmbH, Germany. The authors would like to thank partic-ularly Dr. Karsten Blatter and Miss Maria Strid from theHerberts company for their collaboration.

References

[1] M. Osterhold, F. Niggemann, Prog. Org. Coat. 33 (1998) 55.[2] P. Lange, J. Coat. Technol. 56 (1984) 23.[3] C. Decker, J. Coat. Technol. 59 (1987) 97.[4] C. Decker, K. Moussa, J. Coat. Technol. 62 (1990) 55.[5] M. Koshiba, K.K.S. Hwang, S.K. Foley, D.J. Yarusso, S.L. Cooper,

J. Mater. Sci. 17 (1982) 1447.[6] J.L. Barrett, J. Radiol. Curing 6 (1979) 20.[7] R. Chandra, R.K. Soni, Polym. Int. 31 (1993) 239.[8] D.R. Pemberton, A.F. Johnson, Polymer 25 (1984) 529.[9] S. Radhakrishnan, R.A. Pethrick, J. Appl. Polym. Sci. 51 (1994) 863.

[10] J.E. Dietz, B.J. Elliott, N.A. Peppas, Macromolecules 28 (1994) 5163.[11] B.S. Chiou, S.A. Khan, Macromolecules 30 (1997) 7322.[12] A.Y. Malkin, S.G. Kulichikhin, Adv. Polym. Sci. 101 (1986) 217.[13] K. Te Nijenhuis, Thermoreversible Networks, Springer, Berlin, 1997

(Chapter 1).[14] P.J. Halley, M.E. Mackay, Polym. Eng. Sci. 36 (1996) 593.[15] H.H. Winter, F. Chambon, J. Rheol. 30 (1986) 383.[16] F. Chambon, H.H. Winter, J. Rheol. 31 (1987) 683.[17] S.B. Ross-Murphy, Polymer 33 (1992) 2622.[18] C.Y.M. Tung, P.J. Dynes, J. Appl. Polym. Sci. 27 (1982) 569.[19] K. Te Nijenhuis, Colloid Polym. Sci. 259 (1981) 522.[20] S.A. Khan, R.A. Frantz, I.M. Plitz, RadTech ’92, p. 770.