`
`Colour
`
`Fiker
`
`L
`
`Lens *~
`r "~ -~
`~'"""""'""'00
`
`Polensauon fik11r
`
`I
`
`Figure 2-3 - Densitometer
`
`As in a spectrophotometer, light is reflected from the substrate under test, and the
`reflected light is passed through a lens or prism, and collected on a CCD. The CCD
`applies four filters, to calculate the reflectance m the cyan, magenta, yellow and
`'black' regions of the spectrum. These filters will be applied electronically, and
`densitometers are also available to measure hexachrome and similar colour printing
`systems.
`
`The reflected light is filtered to only analyse light over a certain wavelength band
`Two examples are shown below, for cyan and yellow, Figure 2-4.
`
`•• ,.
`
`1.2
`
`'
`<IJ u
`c ••
`"' t) DO
`Q) ••
`<IJ
`0:: 11
`
`10
`
`I 4
`
`01
`
`- G)
`
`0~--~--~--~------~~====~
`480
`$»
`,.,
`tlO
`Wavelength (run)
`
`Figure 2-4 - Colour filters
`
`The curves shown in each case represent the amount of light reflected from a sample
`at each of the wavelengths indicated. The box indicates the area over which
`reflectance is measured to calculate the density of the colour.
`
`16
`
`FAST FELT 2010 I pg. 31
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`(
`
`(
`
`The amount of light from the illumination source is known (incident light). The
`
`amount of light reflected is also known. Reflectance is thus defined as;
`
`Reflectance = Reflected light
`
`Incident light
`
`Equation 2.1 -Reflectance
`
`Density is defined as the reciprocal of the reflectance, so that as the ink film thickness
`
`increases, and the reflectance decreases, the density increases. A logarithmic scale is
`
`used to improve the closeness of the results between density and ink film thickness.
`
`Thus density is defined as;
`
`Density= Log10 (!/Reflectance)
`
`Equation 2.2 - Density
`
`This work was extended by Murray [19), who defined a technique for the calculation
`
`of the size of dots. By comparing both the solid density (Ds) (the density measured at
`
`100% coverage) and the halftone density (D1) (the density measured at the coverage
`
`under investigation) a relationship between the optical density and the actual printed
`
`coverage was observed to be;
`
`Area=
`
`1-10-0
`'
`1-10- '
`0
`
`Equation 2.3- Murray Davies equation
`
`Close correlation was observed between densities measured with a densitometer and
`
`the fractional dot area (measured using an optical microscope). Comparing prints and
`
`proofs showed higher densities on the prints, caused by spreading of the ink into or
`
`over the substrate - tone gain. It was therefore demonstrated that tone gain may be
`
`17
`
`FAST FELT 2010, pg. 32
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`calculated from the optical densities.2 Yule and Nielson [20] modified the Murray(cid:173)
`Davies equation to include an empirically derived factor to compensate for light
`scattering within the substrate. This attempts to calculate a value for the physical tone
`gain observed in the sample, ie the degree of spreading observed by the ink on the
`substrate, making dots larger than was intended. This method requires a visual
`assessment of a 50% coverage dot, and as such requires a significant degree of user
`intervention and thus an additional level of variability. For the purposes of this work it
`is attempted to minimise user intervention, and to use properly defined techniques. As
`a result, this technique was not used for the calculation of physical dot gain. Image
`processing provides techniques for the calculation of physical dot gain over the range
`of tonal coverage [21].
`
`However, conventional colour measurement techniques were used, in evaluation of
`scumming, and
`these are discussed
`[ 15]. Using optical
`fully
`in Hunt,
`spectrodensitometry to measure colour, both in terms of the printed ink, and in the
`unprinted areas enabled identification of the levels of scumming between the printed
`substrate (in unprinted regions) and the bare, unprinted substrate.
`
`2.2.2 Surface Profiling and Volume measurement
`
`A gravure cylinder is an engraved, etched or ablated surface, and one focus of this
`work was to analyse and measure the engraving details on the cylinder. As such, this
`section will review the methods available for measuring surface topography, with
`particular reference to those methods that are most appropriate. [22]
`
`Many techniques are available for surface profiling, from the most complex (AFM,
`STM etc) to simply dragging a fingernail over a surface. Each method has its own
`advantages and disadvantages, but since the purpose of the use of surface profiling in
`this case is to investigate indentations, intentionally placed into a surface, rather than
`natural indentations and ridges, the deflection that must be measured may be large and
`
`2 Tone gain is defined as the calculated tonal coverage minus the specified coverage. Thus a 50"'o dot,
`measured to be of 58% coverage exhibits 8% tone gain.
`
`18
`
`FAST FELT 2010 I pg. 33
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`the sides may also be relatively steep. As a result, many of the methods traditionally
`used in engineering for surface profiling will not be applicable.
`
`(
`
`Traditionally, stylus measurement teclmiques are used in the analysis of surface
`profiles [23]. In these procedures a stylus is dragged across the surface being analysed
`whilst in contact with the measurement surface. The vertical displacement of the
`stylus is tracked, and a profile is generated. From this data, various rouglmess
`parameters may be calculated, eg. R3 , Rq, Rz, R, etc. [24]. This two-dimensional data,
`although useful for calculating the rouglmess of a surface, is of little use when
`attempting to calculate volumes, which require three-dimensional measurements,
`although a raster scan may be performed
`to produce a
`three-dimensional
`measurement. However, this is not suitable when measuring steep cell sides as under
`these conditions the stylus sides contact the indentation, and not the tip of the stylus.
`
`Several different three-dimensional methods have been proposed for analysis of
`gravure cells, many of which due to the nature of the cells are impossible to use. Kunz
`[25] simply cut cylinders up to examine them under an optical microscope. Due to the
`cost involved in this sort of destructive analysis, little other work has been published
`using this method, although it does allow reasonably detailed analysis. Scanning
`electron microscopy (SEM) was also abandoned due to the necessary destruction of
`the cylinder to obtain suitable sample sizes.
`
`Certain stylus teclmiques are applicable to three-dimensional measurements. Bowen
`et al [26] [27] discuss the uses and limitations of atomic force microscopy. Atomic
`force microscopy (AFM) uses a very fine stylus mounted on the end of a cantilever,
`held close enough to the measurement surface that inter-atomic forces cause the stylus
`to be deflected. This deflection is measured using a laser, reflected off the gold plated
`top surface of the cantilever, and received by a position sensitive detector. This
`deflection is interpreted into a height profile of the measurement area. It is possible to
`operate an AFM in several modes, including contact (where the stylus is held in
`contact with the surface), tapping (where the stylus is briefly brought into contact at
`each measurement point, but is not dragged between them) or non-contact (where the
`atomic forces are measured to keep the stylus and the substrate separate). The stylus is
`moved across the surface of the substrate in a series of linear measurements, which
`
`19
`
`FAST FELT 2010 , pg. 34
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`are all taken to produce a three-dimensional image. It is however, only applicable to
`
`maximum height deviations of±IO!-lm, due to the length of the stylus protrusion from
`
`the cantilever and thus not usable to examine cells, which may be up to I 00!-lm deep.
`
`Scanning tunnelling microscopy (STM) is another probe I stylus type measurement,
`
`where the probe is maintained close to the substrate, with either a constant 'tunnelling
`
`current' being passed between the stylus and the substrate (which must be electrically
`
`conductive) or with the stylus height being maintained, and the current being
`
`measured to infer the distance between the stylus and the substrate. This was excluded
`
`for similar reasons [28], and particularly time- it will require approximately 1 Y2 years
`
`to measure the area of a single large cell, although this will allow calculation of the
`
`position of each atom in the outer surface of the cylinder chrome surface.
`
`White light interferometry [24] is a recent development, and allows a non-contact
`
`examination to be performed. The principles of interferometry are well established,
`
`but the use of a white light source giving a much longer coherence length and the use
`
`of an L VDT controlled head movement allow much larger scan ranges, and with
`
`minor modifications to the optical systems (detailed in chapter 3.2) allow detailed
`
`analysis of the top surface and insides of gravure cells, including measurement of the
`
`depth, volume, open area, width, and length of the cells.
`
`In white light interferometry a beam of white light is split by a beamsplitter, with half
`
`being passed to a reference mirror, and half being passed to the substrate. These
`
`beams are then recombined, and the resultant interference pattern is captured by a
`
`CCD camera, Figure 2-5
`
`20
`
`FAST FELT 2010, pg. 35
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`CCD
`
`c
`
`A
`
`Illumination
`
`source
`
`Figure 2-5 - Interferometer
`
`Light comes from the iJiurnination source, (A) (normally a standard tungsten-halogen
`
`bulb) and is divided into two beams by the beamsplitter, (B). Half the light is sent to
`
`the reference mirror, and half to the substrate. The reflected light from the mirror and
`
`the substrate is then recombined, and passed to the CCD array, (C). The whole
`
`apparatus is then slowly moved closer to the substrate, and images are regularly
`
`captured for analysis by computer.
`
`It is possible to operate most interferometers in two modes, VSI (Vertical Scanning
`
`Interferometry) and PSI (Phase Shifting Interferometry). PSI is the older technique,
`having been used in interferometers since the early 1980s. It uses a series of
`
`measurements, where the apparatus is moved only Y4 of the wavelength of the light
`
`being used between each measurement. From this measurement, the relative height of
`
`each pixel is calculated. Due to this limited number of measurements, the speed of
`
`measurement is significantly higher than when taking VSr measurements. The
`
`limitation of these measurements is in terms of the maximum height differences
`
`which can be measured, 160nrn in PSI mode, SOOJliil in VSI mode. Vertical resolution
`is improved in PSI mode, typically 3A while with VSI measurements a resolution of
`1 run can be achieved with care. The latter is very appropriate in the current
`
`application and so VSI measurements were used throughout this investigation.
`
`21
`
`FAST FELT 2010, pg. 36
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`2.2.3
`
`Ink transfer measurement
`
`Since colour is a function of ink deposit, which is itself a function of print quality it is
`
`necessary to analyse the quantity of ink being transferred. Deposit is determined by
`
`transfer mechanisms and motivates this review.
`
`Kistler and Schweizer [29] describe several teclmiques for film thickness I ink transfer
`
`measurement. Absorption techniques, either using infrared or microwaves may be
`
`used to examine the thickness of transferred films, whereby the absorption of either
`
`infrared light or microwaves may be measured, and a thickness may be inferred from
`
`the consequent attenuation. This teclmique however, requires greater film thicknesses
`
`than those observed in the gravure process due to the accuracy of the system, and the
`
`need for enough of the film to be present to absorb the infrared/microwaves. X-ray
`
`fluorescence measures the emission of X-rays from suitable elements. Disadvantages
`
`of this method include extremely long sampling times, which are required to acquire
`
`accurate mean values, and the necessary use of unrealistic inks containing tracer
`
`elements for testing purposes.
`
`Capacitance gauging is also occasionally used for analysis of thin films. By placing a
`
`flat plate sensor above a grounded target plate, with the substrate between them, the
`
`capacitance of the parallel plates can be measured, and the thickness inferred from the
`
`change in capacitance between the unprinted substrate and a printed substrate with a
`
`known film thickness applied. It is however, impractical to use this teclmique with
`
`films of the thickness used in the gravure process due to the low precision of the
`
`teclmique. Each of these three teclmiques is only able to be used with even films, such
`
`as are obtained at 100% coverage. They are also only able to be used over relatively
`large areas, of the order of a few mm2 at least. Due to the need to measure much
`
`smaller areas for the analysis of individual dots, these methods were not usable.
`
`The Moire technique (also sometimes referred to as 'mechanical interferometry') uses
`
`interference techniques to measure film thickness, however it is limited to use on
`
`films of greater thickness than lO)lm, and only has a vertical resolution of IOJ.lm. By
`
`22
`
`c
`
`(
`
`(
`
`(
`
`FAST FELT 2010, pg. 37
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`superimposing two optical gratings, one undistorted, and one distorted by the film, it
`is possible to infer the thickness of the film. Again, this is only of use with
`significantly larger areas than are available and significantly thicker film deposits.
`
`(
`
`'Profilometers are typically not used for film thickness measurements' ([29] p226,
`7.3.2.1) because traditional profilometers use a stylus that is dragged across the
`surface of a sample. Optical profilometers however, are extremely suited to the
`measurement of films. Two primary techniques are available. The triangulation
`method uses a laser beam, focused on the surface at test and viewed at a fixed angle
`by a position detector, similar to the technique used for height displacement
`measurement as used in an atomic force microscope. As the height of the surface
`changes (i.e. as the thickness of the transferred film changes) the light spot on the
`position detector moves, thus giving a measurement of displacement of the test
`surface, relative to the detector. The method can use extremely high sampling rates
`(up to 40kHz), high resolution, large measurement ranges (up to I Omm) and a
`relatively high working distance (up to Scm). Unfortunately it is only practical to use
`this on thicker films than are transferred in the gravure process. White light
`interferometry [24] however provides a non-contact, non-destructive technique for the
`analysis of films. The teclmique is described in 2.2.2. By varying the optical
`configuration of the interferometer, suitable x-y measurement areas may be defined
`either for the measurement of individual dots (the gravure process not being a contone
`process) or for continuous films, whilst the vertical resolution of the interferometer
`allows for the measurement of the extremely thin films observed. This technique does
`have several drawbacks, primarily those of time and cost. Measurement of film
`thickness takes as little as 30s, but measurement of film thickness where a broken film
`occurs (for example, at the highlight end of the tonal range) takes approximately five
`minutes per sample. This is, however the only process which allows the measurement
`of such broken films (30].
`
`When calculating the quantity of ink released from the cells it is necessary to either
`know the amount of wet ink that is transferred, or to calculate the volume of dry ink,
`and the amount of solvent which is associated with this quantity of solids. Due to the
`extremely fast drying nature of the ink, these wet measurements are not possible, and
`as such dry analysis is required. A precise breakdown of the ink was supplied by the
`
`23
`
`FAST FELT 2010, pg. 38
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`manufacturer, and the quantity of solvent in the wet ink is known precisely, but the
`
`solvent content in the dry ink must still be measured. Traditionally, to identify the
`
`solvent component of a sample, gas-chromatography mass-spectrometry is used [31]
`
`[32] [33]. Thermal desorption heats the sample, driving off the volatile fraction into a
`moving, inert gas stream, and condenses it onto a cold surface [34] [35]. This surface
`
`is then rapidly heated; releasing the volatiles back into the gas stream, and passed
`
`onto the gas chromatograph - mass spectrometer. Clearly the thermal desorber is
`required to ensure that all the volatile components are injected into the gas
`
`chromatograph at the same time, rather than over a period of several seconds if the
`
`entire sample had to be heated. The sample is then passed through a column, which
`separates out the components. This is then passed on to the mass spectrometer, to
`
`identify the components that are included [31] [36). The mass spectrometer uses
`ionisation techniques to break up the component molecules, and then analyses the
`
`components by comparing the break-up pattern with those of known substances. This
`
`is performed automatically. Once the quantity of solvents in the sample has been
`measured and the volume of the dry dots has been calculated, it is possible to
`
`calculated the quantity of solvent originally related to the solids, and thus the volume
`of wet ink which was released from the cell.
`
`Gas chromatography has been used as an analytical tool since approximately 1950,
`although experimental tools have been used since 1940. The sample under test is
`
`injected into a moving stream of gas, which passes through an oven. Inside the oven is
`
`a long coiled tube, which the gas passes through. By the end of the tube the
`components have been separated by elution. They are distinguished by the different
`
`times they take to pass through the column - the retention time.
`
`Finally, the sample is passed to the mass spectrometer. This is the oldest part of the
`
`system, with mass spectrometry having been used in some form since 1898, and in
`
`conjunction with gas chromatographs since 1958, shortly after its introduction. A
`
`mass spectrometer takes the input molecules and bombards them with electrons to
`
`make ions, which spontaneously break up. This break-up pattern is specific to each
`molecule, and so a piece of software may analyse the patterns observed, and identify
`
`the most likely chemicals from the pattern. In addition, the frequency of each ion (or
`abundance) is measured, thus allowing a detailed breakdown of the components fed to
`
`24
`
`FAST FELT 2010 I pg. 39
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`the mass spectrometer. Results are produced as a 'total ion chromatograph' (TIC),
`which details the frequency of each ion, and each fragment identified from the break(cid:173)
`up of the molecules. Automated analysis follows, identifying each component in the
`sample, and its abundance from the TIC.
`
`There is no published literature indicating that this technique may be used on printed
`polymer films for this specific purpose. Similar trials have however, been performed,
`for example Garcia and Sanz [37] who used the procedure to analyse the volatile
`content of Origanum Vulgaris leaves. For the purposes of this research, a new
`technique was developed in conjunction with the manufacturers of the thennal
`desorber - gas chromatograph - mass spectrometer in use. [38]
`
`2.3 Experimental Results
`
`To consider the mechanisms of ink release from gravure cylinders, the review of
`previous experimental work relating to rotogravure printing is discussed in three
`sections. The first details the effects of altering inks, the second of changing
`substrates, and the third examines the effects of cylinder engraving.
`
`2.3.1 Inks
`
`Rotogravure inks are predominately solvent based. They consist of a series of
`components intended to adhere to the substrate, and are approximately 80% solvent
`by mass. Much of the rest of the ink is made up of pigment and binder. Many things
`may be added to the basic solvent blend, including adhesion promoter (itself another
`solvent), nitrocellulose or polyvinyl butyral, surfactants and many other chemicals
`specific to an
`individual
`ink. Changing any one of these components may
`significantly affect the ink, and thus its transfer.
`
`Bohan et al [39) investigated the effects of changing ink viscosity, demonstrating that
`ink viscosity has a large influence on the colour of the final printed copy with A.E
`values of up to six being recorded. These changes are significantly larger than those
`
`25
`
`FAST FELT 2010 I pg. 40
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`necessary to be detected visually, and point to the importance of ink viscosity It was
`
`found that subtle changes in viscosity at low viscosities are far more significant than
`
`(
`
`changes at higher viscosities. These experiments were also repeated, with the same
`
`findings in Bohan et at [40]. Kunz [25] examined the effects of viscosity on print, and
`
`observes that tone gain is significantly increased at low viscosities, while it is reduced
`
`at high viscosities. He also notes the effect of changing viscosity on the filling state of
`
`the cells, where high viscosity inks do not fill gravure cells as completely as low
`
`viscosity inks. The analysis technique used for this is not entirely clear, but it is
`
`known that he used an early interference technique to calculate the deflection between
`
`the top surface of the ink and the top surface of the cylinder.
`
`Further mention is made of ink temperature, and its effect on ink transfer, although
`
`this is primarily due to temperature affecting the viscosity of the inks under study. It
`
`should be noted that the inks Kunz used are of a significantly higher viscosity than
`
`those which would be used on a press today (36s, Zahn 2 - normal operating
`
`viscosities are of the order of 20s, Zahn 2).
`
`Elsayad et a! [ 41 ] investigated the effects of viscosity and resin content, discovering
`
`that the amount of ink transferred by the cylinder increased as the viscosity of the ink
`increased. It was also found that the type of ink resin used had a considerable effect
`
`on the amount of ink transferred. Benkreira et a! [29] [ 42] [ 43] [44) also investigated
`
`the effects of ink, indicating that it has little effect on the release.
`
`Pekarovicova et al [ 45] investigated the use of hot melt inks for gravure printing, but
`
`as their research was limited to inks which at best were several hundred times more
`
`viscous than gravure inks, not solvent based, and limited to draw-down testing of
`
`them, this has little relevance to the gravure process.
`
`2.3.2 Substrates
`
`Gravure printing may be performed on many different substrates, from extremely thin
`
`papers, such as those used in catalogue printing, to heavy papers as used in art books.
`
`From thin packaging films (which may themselves be metallised or coated in some
`
`26
`
`FAST FELT 2010 , pg. 41
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`other fashion) to thick cartonboard as used in cigarette packs. All the work in this
`investigation was performed on plastic substrates, due to the non-porous nature of
`plastic films, but even here, by changing the type of film, the surface roughness and
`specific surface energy may change significantly. The specific surface energy relative
`to the surface tension of the ink may well need to be changed, using (for example)
`corona treatment, to enable the ink to transfer and adhere.
`
`Kunz [25] indicates that the substrate is an extremely significant variable. He
`indicates that there is a "clear connection between smoothness and printability",
`suggesting that there is an operability window for smoothness, where substrates that
`are either too rough, or too smooth will have "negative effects"3
`. It is also observed
`that for printing to be achieved, it is necessary for the impression roller to force the
`substrate into contact with the ink, and that this is "problematic" due to the need to
`squeeze the substrate into the cells, as the meniscus of the ink lowers the top surface
`of the ink below the top surface of the cylinder. However, this work was performed
`using ink significantly more viscous than that used in the modern process (34s Zahn
`2) and at significantly lower speeds than those used on commercial presses.
`
`LaFaye et al (46] produced different papers and printed them on test equipment (some
`were also printed on commercial equipment) to test their printability. They show that
`both solids content and coating weight (both of which promote surface smoothness)
`have a positive effect on printability. This correlates with (25], where it is shown that
`generally smoother substrates print better. Heintze [ 47] also investigated the
`smootlmess of coated board on a GRI printability tester. He discovered that the wear
`of the doctor blade can affect the print quality, particularly in terms of the 'speckle' of
`the print. He acknowledged that the use of a lamella blade minimises this problem.
`
`Hansson and Johansson [ 48] [ 49] examined the printability of different substrates
`using stereo photometric techniques, using irradiance to infer a measurement of
`surface height and roughness. They also determined that areas of substrate
`significantly lower than the average surface tend to have missing dots. This method is
`
`3 The comparison is given to water on glass, which beads up and does not adhere
`
`27
`
`FAST FELT 2010 I pg. 42
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`however, not as accurate as white light interferometry, which was used in the conduct
`ofthis investigation.
`
`2.3.3 Cylinder analysis
`
`Tucker [50] examined the use of interferometry to evaluate anilox rolls, but since an
`anilox roll is an intaglio roller, rotating in an ink bath, the principles of measurement
`hold for measurement of gravure cylinders. Interferometry was considered to be a
`breakthrough in measurement technology, replacing the traditional methods of
`theoretical calculation, and
`liquid volume measurement. Theoretical volume
`measurement is the standard technique where an optical microscope is used to
`calculate the open area at the roller surface, and to infer the volume from these
`physical dimensions. It is limited by the operator's identification of edges, and the
`assumptions that must be made about the shape of the cells, which, for the purposes of
`calculation must be assumed to be ideal, despite the limitations which this imposes.
`Liquid volume measurement is an indirect technique, by which a known volume of
`liquid is applied to the cylinder, doctored, and then 'printed' onto a substrate. The
`area printed can then be calculated, and the volume inferred from this measurement.
`This again, is limited by the operator as the doctoring process is likely to influence the
`area over which the liquid is spread and calculation of the printed area is dependent
`on the users interpretation [51]. It also assumes that the cells are completely filled,
`which has already been shown by Kunz [25] not to be true.
`
`2.3.4
`
`Ink Transfer
`
`Several variables are believed to influence ink transfer. Ofthese, only engraving type
`and screen ruling have been investigated. This section reviews the work that has been
`performed in this area. This section also includes scumming as this can be considered
`as unwanted ink transfer.
`
`Little work has been performed in evaluating the different forms of engraving in
`terms of the ink release. Kunz [25] however completed the earliest work, and he
`
`28
`
`FAST FELT 2010 I pg. 43
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`examined the filled and emptied states of four different types of cells, establishing that
`
`laser engraved cells release the most ink (88%), followed by two types of etching,
`round dot (80%) and conventional (75%), with electromechanically engraved cells
`
`releasing the least ink (68%). This he attributed to the shape of laser engraved cells,
`
`which have very smooth bases and lack the pointed apex of electromechanically
`
`engraved cells. This was, again, established by the use of an early interferometric
`
`microscope, but under conditions far removed from the modem industrial process.
`
`Pulkrabek and Munter [44] examined the release of ink from a knurled roller,
`
`analogous to a gravure cylinder. Their work indicates that 59% of the volume of the
`ink in the roller grooves is transferred to a substrate, but that the release is inversely
`
`proportional to the fluid viscosity to "a small degree". It should be noted however,
`
`that their ink was highly shear thinning, with a rest viscosity of 5000cP
`(approximately 20,000s - Zahn 2) - or approximately 1 OOOx more viscous than
`
`normal process gravure ink, and a thinned viscosity of approximately lOOcP (43s (cid:173)
`Zahn 2) - or approximately twice as viscous as normal gravure ink. No indication is
`given of how the portion released was measured.
`
`Benkreira and Patel [ 42] measured transfer using infrared absorption techniques in
`
`reverse gravure coating. It was discovered that the amount of ink released from a cell
`
`appears to be consistent at approximately 32% of the transferring cell volume. By
`
`using three different types of cell engraving, it was also established that the release is
`independent of the type of cell, except at very low speeds, where variation was
`
`observed. Benkreira et al. [43] measured the transfer from both loaded and unloaded
`
`forward gravure cylinders, establishing that unloaded cylinders release approximately
`20% of their volume, but that loaded cylinders release up to 33% of their volume. It
`
`should be noted that no indication of the impression pressure was given, despite
`
`significant evidence that the impression pressure will affect the release from the
`
`cylinder. Benkreira and Cohu (52] continued this work, demonstrating that with
`
`unloaded forward gravure coating release is of the order of 15-20% of cell volume.
`
`This work also demonstrates that the thickness depends on operating parameters,
`
`including speed and cell geometries. It should be noted that all this work was carried
`
`out on test rigs, not on production scale presses. It should also be noted that the
`infrared analysis equipment used in all three of these studies provides an accuracy of
`
`29
`
`FAST FELT 2010 I pg. 44
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`
`
`(
`
`±0.5um - or approximately 10% of the thinner film thickness measured. The
`technique used for analysis relies on infrared absorption over a large area, and
`averaging over a period of time. As a result, the technique is not applicable to the
`measurements being performed here. Furthermore it should be noted that all this work
`has only been performed on solid coverages, and not the halftones being examined in
`this work.
`
`Jeske [53] simulated the gravure process with an impulse driven press, where a loaded
`mass forced the substrate onto the printing plate. Although this is similar to the
`earliest gravure printing techniques, several problems occur with this technique.
`Firstly, the printing plate forces the substrate into a completely different contact
`regime than a cylinder. Secondly, the plate was doctored by hand, despite Kunz's [25]
`and others assertion that doctoring can, and will affect the filling state, and thus the
`release characteristics of the cells. Despite the criticism that "unrealistic inks" were
`employed in the research of other authors, Jeske used a highly viscous ink, mixed
`with oils to reduce the viscosity, rather than solvents. No indication is given of the
`size of the measurement area, but analysis of transfer was calculated by the
`differential mass method, the accuracy of which is low, particularly for practical
`conditions. It is also known that the surface tension of the ink affects release to some
`extent, as the ink is 'sucked' onto the substrate (Kunz [25]) yet the inks used here,
`have a significantly higher surface tension than inks made with solvents. It is also
`demonstrated here, through the use of photomicrographs that the cells are not
`completely filled, although the technique used to demonstrate the lack of filling does
`not allow quantification of the fill state.
`
`Kapur, Gaskell and Bates [54] investigated the release from a gravure cylinder in an
`offset gravure coating test rig. Tests