Some applications of viscometry from studies on the absorption of vegetable oils and solvents by...

In previous studies on the drying performance of oleoresinous and solvent-based offset lithographic printing inks, a test was devised to determine the rate of absorption of drops of vegetable oils and solvents deposited on the surface of a paper substrate. The influence of viscosity on the rate

of horizontal capillary absorption indicated a potential use of the absorption indicated a potential use of the absorption data to predict viscosity in the oxidative polymerization of polyunsaturated vegetable oils. Volumes as low as 1.0 mL could be used in thermal oxidative polymerization studies and tested by the Drop Deposition method. The stain radii [R.sub.s] of absorbed drops of the oil, e.g., on blotting paper or filter paper substrates, follow a linear relationship with the square root of time ([square root of (t)]s) and the slopes m of the [R.sub.s] vs [square root of (t)] graphs are converted to dynamic viscosity in dPa.s units using a modified form of the Lucas Washburn equation from which m [alpha] [square root of (1/2[eta])]. The study has implications for assessing the stability of vegetable oils used in frying operations. The test method may also be used in the study of room temperature oxidative polymerization of polyunsaturated vegetable oils to evaluate drier catalysts used in surface coatings.

Keywords Drop Deposition Method, Polyunsaturated vegetable oils on paper, Capillary absorption, Viscosity prediction, Thermal oxidative polymerization, Frying oils stability, Room-temperature oxidative polymerization

Symbols and abbreviations

h height of rise of liquid (cm) in time t (s).

r mean pore radius (cm)

[theta] contact angle (liquid to paper)

[gamma] surface tension of the liquid (mN [m.sup.-1])

[eta] viscosity of the liquid deci Pascal

seconds (dPa.s)

[R.sub.s] stain radii (cm)

r mean pore radii (cm)

D diameter (cm)

CD cross direction (referring to paper)

MD machine direction (referring to paper)

m the slope of the [R.sub.s] vs [square root of (t)] graph

[C.sub.i] intercept on the [R.sub.s] axis

V (or vol.) volume (mL)

t time

[square root of (t)] square root of time (in seconds unless another time period is stipulated)

g/[m.sup.2] grams per square meter (referring to paper)

[mu]m microns

ssa specific surface area

K value a rate of oxidative thermal polymerization derived from K = [[[log.sub.10][[eta].sub.2] - [log.sub.10][[eta].sub.1]]/[[t.sub.2] - [t.sub.1]]]

HBPF high boiling petroleum distillate solvents

SFO sunflower seed oil

ARLO refined linseed oil

ROO refined olive oil

EVOO extra virgin olive oil

TOFA-Me. Ester (Tall Oil Fatty Acid Ester)

DD Drop Deposition

PRA Paint Research Association

MDF Medium Density Fiberboard

Introduction

This study extends the work previously reported (1.2) to gather information on the drying performance of offset lithographic inks printed on various coated and uncoated paper substrates. One objective was to determine the effect of substituting the distillate component of an oleoresinous and solvent-based ink with a vegetable oil. A test called the Drop Deposition Method was devised to determine the effect on the setting speed of the ink film during the capillary absorption stage of drying which takes place immediately after the printing impression (the pressure penetration stage) and subsequently when the printed sheet enters the pile (stack of printed sheets). Attention may be drawn to some early studies concerning capillarity and wetting in paper structures, (3) the penetration of varnishes and inks into paper under pressure, (4) and the physical surface properties of papers and their relevance to printability. (5) Later work (6.7) involving interaction of liquid drops on porous substrates has been informative for the present study.

The influence of the viscosity of the oil or solvent on the rate of absorption indicated the possibility of using absorption data to predict viscosity in the oxidative polymerization of polyunsaturated vegetable oils using volumes as low as 1.0 mL for the investigation.

Experimental

Reviewing earlier studies, capillary rise experiments, following the work of Tollenaar, (3) were carried out on a range of papers and various vegetable oils, methyl fatty acid esters, and high boiling petroleum distillate (HBPF) solvents.

From a simplified form of the Lucas Washburn equation:

h = [square root of ([[r[gamma]cos[theta]t]/[2[eta]]]

where h is the height of rise of liquid (cm) in time t (s) (see Fig. 1); r is the mean pore radius (cm); [theta] is the contact angle (liquid to paper); [gamma] is the surface tension of the liquid (mN [m.sup.-1]); [eta] is the viscosity of the liquid (dPa.s); and t is time (s).

[FIGURE 1 OMITTED]

For a given substrate and liquid

h [infinity] [square root of(t)]

Drop deposition method

Drops of the liquid (0.01-0.05 mL) are applied to the surface of the substrate by means of a micro syringe. The absorption of each drop is timed with a stopwatch to the point of surface liquid disappearance. Coincident with this end-point, the periphery of the stain is marked with a pen point and the diameters are measured in the machine and cross directions of the paper substrate. The stain radius [R.sub.s] is calculated from the geometrical mean [square root of([R.sub.M] X [R.sub.c])] of the two radii. This is because the stains show ellipticity, more evident as the drop sizes increase. See Fig. 2.

[FIGURE 2 OMITTED]

The elliptical tendency is due to the anisotropic properties of the substrate, capillary absorption properties differing between machine and cross directions of the paper thus [R.sub.M] > [ R.sub.C]. Using the data from the absorption of solvents and oils on cartridge paper, the graph [R.sub.s] vs [square root of(t)] is a straight line. Having established this linearity it was considered no longer essential to apply drops of measured volume, thus simplifying the procedure. Drops could be deposited from the tips of wire or glass rods of narrow diameter using a spread of drop sizes approximating to the range 0.01-0.05 mL. The relationship of [R.sub.s] vs [square root of(t)] is linear for cartridge paper (150 g/[m.sup.2]) and various polyunsaturated vegetable oils, methyl fatty acid esters, a machine oil, and hydrocarbon solvents (Fig 3). For coated papers, the relationship is a curve of increasing slope but log [R.sub.s] vs log t is linear (Fig. 4).

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

For uncoated paper, [R.sub.s] = m[square root of(t)] + [C.sub.i] where m is the slope of the [R.sub.s] vs [square root of(t)] graph and [C.sub.i] is the intercept on the [R.sub.s] axis.

Using a modified form of the Lucas Washburn equation

[[[R.sub.s] - [C.sub.i]]/[square root of (t)]] = m = [square root of ([[r[gamma]cos[theta]]/[2[eta]]]]

For a given substrate with average pore radius of r (cm), the properties of the liquid to be considered in the absorption process are the contact angle [theta] the liquid makes with the paper surface, the surface tension of the liquid [gamma], and the viscosity of the liquid [eta]. [R.sub.s] vs [square root of(t)] has been established for a range of vegetable oils, methyl fatty acid esters, and high boiling petroleum distillate solvents tested with blotting paper, newsprint, and filter papers of varying porosity (Fig. 5). A feature of the graphs is the marked difference between the slopes of the absorption regression lines for the solvents compared with those of the vegetable oils. A major factor is the viscosity of the absorbing liquid. This is to be noted from the slopes m in Figs. 3 and 5. Whereas the vegetable oils listed have a viscosity some ten times that of the high boiling petroleum distillate, the methyl fatty acid derivatives are much closer, and in fact have absorption characteristics approaching those of the of the 280/310 (HBPF) solvent.

[FIGURE 5 OMITTED]

This is a useful property when formulating inks in which the solvent is being substituted with a vegetable oil derivative. Using absorption data to predict viscosity, graphs (Fig. 6) are constructed from the [R.sub.s] vs [square root of(t)] data for each of the substrates tested with odorless kerosene, methyl fatty esters, sunflower seed oil (SFO), and castor oil BP grade.

[FIGURE 6 OMITTED]

The slopes m from the equation [R.sub.s] = m[square root of(t)] + [C.sub.i] are plotted vs [square root of([1/[2[eta]]])] and [square root of([[gamma]/2n]])].

Table 1 lists the viscosity and surface tension data of the test liquids. The curves of Fig. 6 show the linearity of m vs [square root of ([gamma]/2[eta])] which would be the general equation, but m vs [square root of ([1/[2[eta]]])] is used because the surface tensions of the oils are very similar. When taking samples to monitor the reaction only small drops are needed for the absorption tests. Note here that [R.sub.s] vs [square root of (t)] absorption tests are conducted at a room temperature of 20[degrees]C and the viscosity of the liquids so determined are in dPa.s at 20[degrees]C. Using viscosity vs temperature data for a particular oil or solvent enables an adjustment to be made in the DD test if the temperature is [+ or -]5[degrees]C of the 20[degrees]C measurement. Alternatively, carrying out a control test with the original unexposed oil along with the reacted oil using the same specimen of filter paper allows corrections to be made.

Table 1: Properties of solvents, ester, and oils

                           Viscosity         Surface tension
                      dPa.s @ 20[degrees] C  mN/[m.sup.-1] @
                                               20[degrees]C

Soybean oil                 0.64                  33.0

Sunflower seed oil          0.60                  33.0

Rapeseed oil                0.66                  32.0

Linseed oil                 0.47                  35.0

Mineral oil,                0.21                  33.0
(Singer machine oil)

White spirit                0.01                  23.0

Odorless kerosene           0.02                  27.0

280/310 HBPF (a)            0.06                  28.0

Methyl fatty acid           0.075                 32.0
 esters type A(b)

Methyl fatty acid           0.08                  32.0
 ester of tall
oil F.A (c)

(a) 280/310 High boiling petroleum distillate--low aromatic
content (Haltermann)
(b) Methyl fatty acid ester: A mixture of unsaturated and saturated
fatty acid esters. Mixture of C 12 to C18 chain length. Arco Chemical
(c) Nirez 9011 Arizona Chemicals

The effect of contact angle and surface tension parameters

Reference to previous work on surface absorption of liquids by porous substrates is noteworthy. In a study on the forced penetration and capillary absorption of a range of mixed litho varnishes into super calendared paper, (5) relevant to our work is the finding that in the capillary absorption there is a liner relationship between the rate of (penetration) (2) ([mu][m.sup.2]/s]) and the ratio of surface tension and viscosity [gamma]/[eta].

The results were considered to indicate a zero contact angle (cos [theta] =1) although the surface tension was practically the same for six litho varnishes tested. Hence the graph could be regarded as showing the effect of viscosity. In the capillary rise experimens, (1) it has been possible to get information on rcos [theta] from the slopes of the h/[square root of (t)] graphs. Tollenaar (3) deseribes the factor rcos [theta] in the Lucas Washburn equation as the "effective pore radius" of the paper. He states that the consistent measurements of the effective pore radius from the ascent of different organic liquids on one paper prove that cos [theta] is a constant and since it is improbable that the organic liquids from paraffins to glycol are giving the same finite contact angle, the only acceptable conclusion is that for these organic liquids the contact angle takes very low values. Table 2 lists the properties of the four test liquids and blotting paper (150 g/[m.sup.2]) substrate. It is significant that at constant value r, cos [theta] values for all four liquids are very close, which leads us to the same assumption as that of Tollenaar, that the contact angles must be very low. There are difficulties in assigning accurate values for contact angles of liquid drops applied to paper substrates. The need to use extremely minute drops was emphasized (8) to minimize the effects of gravitation which tends to flatten the drop. From studies (6) on the interaction of the ink jet droplets on porous substrates it has been stated that with rough and porous papers the contact angle is highly time dependent and that surface fiber and void structure morphology and bulk porous structure all contribute to the contact angle. A model (7) to describe the spreading and imbibition of liquid droplets on porous substrates is proposed to be of relevance to the interaction with porous substrates, such as paper. The variation of contact angle with time is demonstrated with droplets of an aqueous solution of glycerine and hexylene glycol deposited on micro porous membranes showing quite marked decreases as spreading and imbibition proceeds.

Table 2: Effective pore radius

Test          [eta] @ 20     [gamma]      h/[square      r cos
oil/solvent   [degrees]C,  mN[m.sup.-1]    root of     [theta] [mu]
              dPa.s        @ 20[degrees]   (t)]       [10.sup.-4] cm
                                  C       Capillary
                                          rise rate

Odorless         0.02           27           0.26          1.0
kerosene

280/310 HBPF     0.064          28           0.153         1.1

Me ester A       0.075          32           0.153         1.1

Soybean oil      0.64           33           0.054         1.1

Evaluation of r cos[theta] from
h = [square root of ([[r[gamma]cos[theta]t]/[2[eta]])]
r cos [theta] = [(h/[square root of (t)]).sup.2] x [2[eta]/[gamma]]

It has not been possible to get consistent results for the apparent contact angle measured immediately after deposition of the drop on the substrate, especially with the more absorbent blotting paper and some filter papers because of the rapid onset of capillary absorption affecting the contact angle, as the volume of the drop on the surface decreases. Figure 7 illustrates a technique developed to determine the apparent contact angle at the instance of deposition of a liquid drop of known volume on the surface of a filter paper. The horizontal capillary absorption is timed from deposition to final disappearance of surface liquid, taking periodic measurements of the diameters [D.sub.M] and [D.sub.C] in the machine and cross direction of the paper. These diameters transverse the contact area of the drop and the surrounding stain area as illustrated. The combination of Whatman WH542 filter paper and SFO gives a sufficiently long absorption time at each measurement. Figure 8 is a graph of the radii [r.sub.M] and [r.sub.C] vs [square root of (t)] according to the straight line equation, [r.sub.M,C] = m[square root of (t)] + [C.sub.i]. The intercept [C.sub.i] at time t = 0 gives an approximate measure of the contact radii of the drop. Given [r.sub.M] and [r.sub.C] the geometric mean [square root of ([r.sub.M] x [r.sub.C])] should be a close estimate of the actual contact radius of the drop. With a drop of known volume V and contact radius r, we have the data to determine h the height of the drop from the equation 6V = [pi]h (3[r.sup.2] + [h.sup.2]) and from this equation the contact angle [theta] from h/r = tan([theta]/2). (9), (10)

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

Cos [theta] at time t = 0 determined for filter paper WH542 and for several liquids is given below:

Table 3 compares the regression equations for the relationships, m vs [square root of ([1/[2[eta]]])], [square root of ([gamma]/[2[eta]])], [square root of ([[[gamma]cos[theta]]/[2[eta]])].

Table 3: Comparison of m vs [square root of (1/[2[eta]])],
[square root of([gamma]/[2[eta]])],
[square root of([gamma]cos[theta]/[2[eta]])]

Oil/solvent              m     [square root of  [square root of
                                (1/[2[eta]])]   ([gamma]/[2[eta]])]

SFO                    0.0291      0.9098             5.2270

PRA 276 oil            0.0134      0.4256             2.3960

Castor oil/Me. Ester   0.0162      0.4975             2.7250
TOFA blend

Me Ester TOFA          0.0651      2.0400            11.5400

Odorless kerosene      0.1440      5.0000            25.9800

Me. Ester A            0.0800      2.7110            15.3390

Oil/solvent        [square root of                   cos      [eta]
                   ([[[gamma]cos[theta]]/[2[eta]])]  [theta]  20 dPa.s

SFO                     5.2160                        0.9961     0.604

PRA 276 oil             2.3760                        0.9832     2.760

Castor oil/Me.            -                             -        2.020
Ester TOFA blend

Me Ester TOFA          11.5300                        0.9972     0.1200

Odorless kerosene      25.9200                        0.9950     0.0200

Me. Ester A            15.3160                        0.9970     0.0700

Substrate: Whatman WH542 filter paper, DD tests @ 20[degrees]C
Regression equations and regression coefficients

a = 0.02846 0.00546 0.00549
b = 0.00301 0.00041 -0.00005
[gamma] = 0.9991 0.9990 0.9989
[[gamma].sup.2] = 0.9982 0.9980 0.9978

SFO            0.9961
PRA 2.76 Oil   0.9832
Me Ester TOFA  0.9972

These are shown to illustrate the good correlation for the oils and solvents tested.

Calibration of a substrate with viscous oils

PRA standard viscosity oils in the range 2-65 dPa.s at 20[degrees]C have been tested by the DD method on Whatman WH41 filter paper. The m values from the [R.sub.s] vs [square root of (t)] are plotted against [square root of ([1/[2[eta]])] (Fig. 9).

[FIGURE 9 OMITTED]

The linear relationship indicated the utility of the test for investigating the thermal oxidation of unsaturated vegetable oils. For this purpose, 20 mL samples of sunflower seed oil (SFO) have been exposed to a temperature of 200[degrees]C in a domestic gas oven for increasing periods of time. The oils of viscosity 1-132 dPa.s, as determined by efflux and falling sphere viscometric methods, have been used as standards to calibrate the filter paper viscometer. The same specimen of filter paper is used for the DD absorption test on the polymerized SFO oils. Any difference in porosity in the same type of filter paper calls for a separate calibration.

Applications for a paper viscometer

The advantage of using minute amounts of oil for the DD test to predict viscosity means that quantities as low as 1.0 mL can be exposed in a thermal oxidation experiment and the reaction monitored by means of a viscosity vs time of heating curve. A practical application could be to determine the stability of an unsaturated vegetable oil in a frying of food operation. As the viscosity of frying oil was found to increase as it deteriorated due to oxidation and polymerization of the component fatty acids, instruments for determining the viscosity of a hot frying oil during use have been investigated. (11) Viscosity prediction from an absorption test on a sample taken from the frying oil vessel and cooled to room temperature (20[degrees]C) before the DD tests are carried out is another simple way of evaluating the viscosity increases in dPa.s units and should relate to the trend with other established tests including the determination of polar compounds, the oxidized fatty acid components, and polymer content.

Another potential application could be in the studies of the ambient temperature oxidation of polyunsaturated oils involving metallic drier catalysis.

Experimental procedure, thermal oxidation studies

Volumes from 1 to 20 mL of vegetable oils containing varying amounts of constituent polyunsaturated and mono-unsaturated fatty acids in glycerides (Table 4) were heated in stainless steel, tin-coated steel, and aluminum dishes (dimensions in Fig. 10) on the top shelf of a Cannon domestic gas oven. No fan assistance was used; so to account for the temperature gradients in the oven, an oven dial thermometer was placed as close as possible to the dish and checked periodically to ensure that the recorded temperature was as near as possible constant throughout the exposure. After a period of 0.5, 1, 2, 3, etc. hours, the dish was removed, cooled to room temperature, and drops of oil were tested by the DD method using the same specimen of paper throughout the experiment. From the absorption data for each example, the m values from the [R.sub.s] VS [[square root of(t)] graphs were converted to dynamic viscosity units (poise) at 20[degrees]C from a calibration graph previously compiled for the particular substrate using the standard viscosity-polymerized SFO oils (Fig. 11). Assuming cos [theta] = 1 and negligible changes in [gamma], the following expression can be used:

Table 4: Characteristics of vegetable oils (% component of fatty acids
(a))

Oils      Linolenic  Linoleic  Oleic  Saturated   Calculated
                                                 iodine value (b)

SFO A         -         65       23      12          132
SFO B         -         66       18      16          130
SFO C         -         63       21      16          127
ARLO (c)     53         15       20      12          182
ROO           -         10       72      18           79
EVOO          -          9       76      15           81

(a) Calculated from data on bottle label
(b) Grams [l.sub.2]/100 grams oil
(c) Typical

Key: SFO, Sunflower seed oil; ARLO, refined linseed oil;
ROO, refined olive oil; EVOO, extra virgin olive oil

[FIGURE 10 OMITTED]

[FIGURE 11 OMITTED]

[[([m.sub.1]).sup.2]/[([m.sub.2]).sup.2]]X[[eta].sub.1] = [[eta].sub.2]

where [m.sub.1] is the slope of the [R.sub.s] vs [[square root of(t)] graph for oil of viscosity [[eta].sub.1] and [[mu].sub.2] is the slope of the [R.sub.s] vs [[square root of(t)] graph for oil of viscosity [[eta].sub.2].

The predicted values are within [+ or -]5% of the viscosities determined by instrumental methods. Heat up time after replacing the dishes of oil in the oven after completion of each absorption test was quite rapid, as confirmed by taking the temperature with an infrared thermometer. Figure 12 illustrates the trend of oxidative polymerization of 20 mL SFO heated at gas mark 4 (actual temperature 180[degrees]C) in dish F for a total of 4 h. The DD test slope lines and converted units in dPa.s are shown for the substrate, cartridge paper (150 g/[m.sup.2]). The calculated dynamic viscosity for the sample at 1, 2, 3, and 4 h is plotted on a log viscosity vs time of heating at 180[degrees]C basis. The loss of weight after heat treatment was 1% on the original weight of the oil.

[FIGURE 12 OMITTED]

The specific surface area

The value a/v in the table of Fig. 12 is known as the specific surface area (ssa). In frying experiments, it is defined as the ratio of the interfacial area between the air and the frying oil and the volume of the oil.(12) This is significant with regard to the rate of oxidative polymerization as determined by the rate of viscosity increase. Hence it is necessary to consider loss of weight due to volatilization of oxidative degradation products (fatty acids, aldehydes, hydrocarbons, alcohols, etc), as well as the effects of exposing the different volumes of oil in dishes of varying diameter.

Calculation of K value

The K value shown in Fig. 12 is calculated from a relationship proposed by Cannegieter (13) for constant high-temperature thermal polymerization of drying oils.

K = [[[[log.sub.10][[eta].sub.2]] - [[log .sub.10][[eta].sub.1]]]/[[t.sub.2] - [t.sub.1]]]

where [[eta].sub.2] and [[eta].sub.1] are the viscosities in dPa.s determined at 20[degree]C at times [t.sub.2] and [t.sub.1] (min). This relationship holds for the slopes of linear curves in the high-temperature (e.g., 280[degrees]C) polymerization of drying and semi-drying vegetable oils. (14) In the process control of stand oil manufacture, samples are drawn from the reaction vessel and cooled to a temperature of 20[degree]C for viscosity determination. Viscosity at any later stage is predicted by extrapolation of the linear log vs t curves. The K values shown on the graphs are all measured at 20[degrees]C and when comparing different oils the same reaction temperature is used.

The graph, Fig. 12, is linear for the 4 h of heating at 180[degrees]C. In experiments where there has been an increase in the slope of the log viscosity vs time curve, an overall K value is recorded.

K = [[[log.sub.10][[eta].sub.2](final product)] - [[log.sub.10][[eta].sub.1](unheated oil)]/Time(min)]

Figure 13 compares the thermal oxidation in dish F of two 20-mL samples of SFO referenced A and B, temperature setting Gas Mark 4 ([180[degree]C). Sample A was taken from a half-filled 1-L plastic bottle and sample B from a full 1-L bottle. Although the total unsaturation (di and mono) of the two oils is similar, the K value for A is 2.5 times that of B. This could be due to more peroxide development in the unheated sample of A due to its storage for several weeks with an air gap over the surface; whereas sample B was taken from a full bottle stored over much the same period. Figure 14 compares the thermal oxidation of samples A and B at the higher oven temperature, gas mark 6 (200[degrees]C) together with a third samples C from a full 1-L plastic bottle tested only a few days from purchase so that it was the least likely to show any signs of peroxide formation. Although no quantitative evaluation for peroxide value has been carried out on the parent oils, a colorimetric test used to identify presence of hydroperoxide groups (OOH) was adopted. Next, 0.1 mL of a solution of cobalt octoate drier (cobalt metal content 0.01 g/mL) was added to 3.5 mL of oils A, B, and C. After stirring, drops of the treated oils were spotted on a white tile and the color changes from the initial pale yellow were observed.

[FIGURE 13 OMITTED]

[FIGURE 14 OMITTED]

                    Oil A         Oil B    Oil C

Trend line      A1    A2    A3   B1    B2   C
Dish            D     E     F    E     F    F
Vol(mL)         1     6.5   10   6.5   10   20
a/v(ssa)        9.6   11.6  7.8  11.6  7.8  3.9
Paper           BP    BP    BP   BP    BP   WH41
K x [10.sup.3]  12.8  6     4.4  5     3.6  3.2

Oil A--Deep green

Oil B--Light green

Oil C--Pale violet

The presence of OOH groups in the parent oils is detected by the oxidation of divalent [Co.sub.2+] (violet) to trivalent [Co.sup.3+] (green).

The test indicated that oils A and B contained OOH groups at a level sufficient to effect the color change; whereas oil C, because it was a fresh oil, retained the pale violet color of the added cobalt drier solution. The deep green of sample A is most likely indicative of some oxidation due to storage in a half-filled bottle with an air gap over the surface of the oil--a theory advanced for the high K value in the thermal oxidation. Figure 14 shows the effect of heating different volumes of oil in various sized dishes. The high K value of oil A heated in dish D is no doubt due to the high ssa value and very low volume of oil. Here, there was a 10% weight loss after heating so that the actual ssa would have been even higher than the recorded figure. At the higher reaction temperature, sample A continues to exhibit a higher K value than that of sample B, but the difference is less marked. Oil C has the least polyunsaturation of the 3 oils (Table 4), hence the expected lowest K value as fresh oil, but here the comparatively lower ssa value may have more significance. Figure 15 compares the thermal oxidation of SFO sample B with a sample of refined olive oil. Typically the latter oil contains a lower amount of constituent polyunsaturated fatty acids. The graph shows the lower polymerization rate of olive oil. Table 5 compares the thermal oxidation rate of SFO sample B and a sample of Extra Virgin olive oil taken from bottles on the day of purchase immediately after opening the bottles. Volumes of 0.25, 0.5, 0.75, and 1.0 mL were heated in aluminum dishes of 2.54 cm diameter giving a spread of ssa values under the conditions shown. The trend of K value with increasing ssa may be noted. Unexpectedly, the lowest volume of olive oil has a higher K value than its SFO counterpart; whereas the other three volumes of olive oil have correspondingly lower K values reflecting the lower proportion of di-unsaturated fatty acids in the glycerides. Possibly at the higher rate of oxygen uptake (more oxygen per unit volume of oil) any accompanying oxidative degradation reactions may have increased ssa, more so in the case of the olive oil heat treatment than with its SFO counterpart. Differences in the non-glyceride constituents (types and content) between the two oils may be significant. Figure 16 is a graph of K value vs the reciprocal of ssa which is v/a equivalent to the depth of oil in a cylindrical vessel, or film thickness for a low volume of oil. This graph shows the trend of decreasing K value with increasing film thickness.

Table 5: Thermal oxidation of sunflower seed oil and extra virgin olive
oil at different ssa values

Oil                            SFO B                    EVOO

Volume (mL)           0.25   0.5  0.75  1.0   0.25   0.5  0.75  1.0

ssa (a/v)            20.3   10.1  6.8   5.1  20.3   10.1  6.8   5.1

Film thickness (cm)   0.05   0.1  0.15  0.2   0.05   0.1  0.15  0.2

[[eta].sup.a]        29.8   11.3  5.2   2.6  53.8    8.5  2.8   1.8
20[degrees]C,
2.5 h,193[degrees]C

K x [10.sup.3]       11.5    8.7  6.5   4.4  12.7    7.4  4.1   2.9

Diameter of dish 2.54 cm
Gas mark 5 (193[degrees])C
DD Test Whatman WH41 filter paper

(a) [R.sub.s] Vs [square root of(t)] 20[degrees]C predicted
viscosity dPa.s

[FIGURE 15 OMITTED]

[FIGURE 16 OMITTED]

Comparison of methods to monitor viscosity increases in thermal oxidation studies

Table 6 compares the viscosities of an 80 mL sample of SFO heated for 4 h at gas mark 5 (193[degrees]C) then tested by the DD method on Whatman 41 filter paper. The predicted viscosity calculated from m vs [square root of(1/[2[eta]])] graph is compared with the viscosity determined by the Zahn Cup no. 2 instrument and the Davison Flow Plate Viscometer. (13), (14) The Zahn cup was calibrated with standard viscosity oils to convert Zahn no. 2 seconds to dPa.s units. The Davison Flow Plate Viscometer was used with a metered volume of 0.5 mL in the inkwells of the flow plate.

Table 6: Comparison of methods to monitor viscosity increase in thermal
oxidation studies

Method                             Original    4 h @ 193     K value x
                                     SFO     [degrees]C (a)  [10.sup.3]

Prediction from DD                  0.60      1.1              1.09
[R.sub.s] vs [square root of (t)]
Zahn cup No. 2 (b)                  0.65      1.2              1.11
Davison flow plate viscometer (c)   0.70      1.3              1.12

(a) 80 mL SFO B heated at Gas MArk 5 (193[degrees]C) for 4 h Dich F
(base diameter 10 cm) ssa = 0.946
(b) Zhan cup seconds converted to dPa.s from:
log [eta] (dPa.s) = (log 1.24 x (zhan cup seconds)) - 1.9759
(c) For Davison Flow Plate vidcometer
log [gamma] = 2.3492 log x + 4.9165
(measured volume 0.5 mL in ink well. See reference 15)

The predicted viscosity is quite close to that determined by the instrumental methods. Moreover, the DD method requires only very small quantities of oil to be taken from the reaction vessel in the thermal oxidation experiment, minimizing any marked changes in ssa. The Davison Flow Plate Viscometer could be useful for determining the viscosity of oil during heating in an industrial frying vessel. A sample could be drawn from the vessel, cooled to 20[degrees]C and tested along with the original unheated oil as the control. A larger sample would be needed for the Zahn Cup, sufficient for its complete immersion.

Validity of K value for determination of thermal oxidative polymerization

In contrast to the high-temperature thermal polymerization of polyunsaturated vegetable oils in the absence of air, or under reduced air pressure (vacuum conditions), or inert gas blanket, thermal oxidation involves oxidative degradation of unsaturated fatty acid chains as well as their interlinking in polymerization reactions. The K value is a measure of the latter. So, for more information on the reaction of both chain degradation and formation of dimers and higher polymers, a more detailed chemical analysis of the products is required.

Factors affecting thermal oxidation

The absence of fan assistance in the gas oven raises the question of the influence of stagnant air layers over the surface of the oil leading to mass transfer problems, all of which could influence the rate of oxidation of the oil.

Oxidation at lower temperatures

The use of the DD method has been investigated to monitor the viscosity increases when small amounts of polyunsaturated vegetable oils were exposed to the atmosphere at temperatures of 18-20[degrees]C and in a plant incubator at 43[degrees]C. Cobalt octoate drier was added to each of the oils at a concentration of 0.1% cobalt (as metal) on the weight of oil. Figure 17 compares the room-temperature oxidation of SFO sample C in dishes D and H and here the effect of ssa is to be noted. The onset of skin formation is presumably at a viscosity at which there is a sufficiency of interlinked triglyceride molecules for a macromolecular structure to be formed. Gel formation occurs soon afterwards in the case of oil with the higher ssa.

[FIGURE 17 OMITTED]

Comparative oxidation with a drying oil

The "linolenic"-rich linseed oil is a drying oil; whereas SFO, a "linolenic"-rich oil of lower iodine value (Table 4), is a semi drying oil with much inferior film-forming properties. Figure 18 compares the room-temperature oxidation of alkali-refined linseed oil and 3 samples A, B, and C of SFO. In contrast to the results of the thermal oxidation when SFO sample A has a higher K value than SFO sample B, the reverse is the case in the catalyzed room-temperature oxidation. No precise explanation for this anomaly is offered, but variations in the degree of chain polymerization and chain degradation reactions, as between the thermally polymerized oils and the metal catalyzed autooxidation at 18-20[degrees]C, may have some significance. The expected higher reactivity of the linseed oil compared with that of samples A and B at equal ssa is demonstrated, the K value for sample L, being more than double that of samples A and B.

[FIGURE 18 OMITTED]

Oxidation of SFO at 43[degrees]C

Ten mL of SFO containing 0.01% cobalt drier was exposed to air controlled at a temperature of 43[degrees]C in an incubator. The trend of oxidative polymerization is illustrated in Fig. 19. The K value does not vary significantly until about 45 h and 10 dPa.s viscosity. K value falls in the next period of 10 h then increases markedly, the oil showing signs of incipient gelation. The DD tests were conducted after first stirring the oil in the dish dispersing any apparent gel structure. Presumably the capillary forces in the DD test are strong enough to prevent gel reformation in the pores of the filter paper. The apparent viscosities after the 58-h period would appear to indicate high polymer formation on the surface of the oil with underlying low-viscosity, less reacted material.

[FIGURE 19 OMITTED]

With reference to the terms "Edge Skin" and "Surface Skin" shown in Figs. 18 and 19, edge skin is detected visually as appearance of fine particles of gel material on the periphery of the surface at the vessel wall. Surface skin means a very fine layer of gelled material forming over the surface, identifiable by lightly probing it with the tip of a glass rod.

A thin film of the SFO with drier content 0.05% (cobalt metal), typical of a varnish film thickness dried in 1.5 h at a temperature of 43[degrees]C, but the film was soft and tacky, remaining so for several weeks. Table 7 compares K values for a fresh sample of SFO treated with three different levels of cobalt drier in dishes giving similar ssa values but lower film thickness than in the case of the previous exposures in the incubator. The increase in K with the content of cobalt drier catalyst is an expected trend. Investigating the DD technique in room-temperature oxidation of polyunsaturated drying oils at film thicknesses comparable to surface coatings (varnishes and paints) may be useful in the evaluation of drier catalysts.

Table 7: K value SFO with cobalt drier variation (exposure at
43[degrees]C)

Cobalt drier  Volume   Dish type    Dish     ssa    Gelation      K x
% (as metal)  (mL)               diameter  (a/v)  time (h)  [10.sup.3]
                                    (cm)

0.10           1.0    Aluminum       5.0    19.63    5.25       3.7

0.05           3.0    Stainless     10.0    26.2     9.5        2.8
                      steel

0.05           1.0    Stainless      5.0    19.63    8.5        2.6
                      steel

0.01           3.0    Stainless     10.0    26.2    12.5        1.7

0.01           1.0    Aluminum       5.0    19.63   17.0        1.5

0.01           2.0    Aluminum       5.0     9.8    17.0        1.3

DD test substrate Whatman LH41 filter paper

Conclusions

Predicting viscosity from absorption data has been possible using a range of papers of varying porosity, Whatman WH41 filter paper conveniently covering the range of 0.5-100 dPa.s.

By using oil absorption data to predict viscosity it has been possible to monitor the thermal oxidative polymerization of polyunsaturated vegetable oils using volumes as low as 1.0 mL for the reaction.

This has implications for assessing the stability of polyunsaturated vegetable oils used in frying operations and further work is proposed to establish relationships with tests such as determination of polar compounds, oxidized fatty acid components, and polymer content.

Using the DD test to predict viscosity in the auto-oxidative polymerization of polyunsaturated vegetable oils exposed at ambient temperatures may provide information on the performance of the oils in surface coating media. It could be of continual use for the evaluation of printing inks and coatings deposited on any surface, which is permeable to a degree that would influence film formation and final drying. A future investigation is planned for the potential use of the technique to study the receptivity of coatings applied to rigid boards such as MDF.

Acknowledgment The authors would like to thank Roger Wallis of Wallis Surface Science for his assistance in the preparation of this paper.

References

(1.) Hutchinson, GH, "A Review of Progress in the Technology and Applications of Offset Lithographic Inks." Surf. Coat Int., 78 (5) 188-194 (1995)

(2.) Hutchinson, GH, "Ink Technology Past, Present and Future." Surf. Coat. Int., 85 (B3) 169-176 (2002)

(3.) Tollenaar, D, "Capillarity and Wetting Properties in Paper Structures. Properties of Porous Systems." Symposium, Surfaces and Coatings Related to Paper and Wood. pp. 195-219. State University College of Forestry. Syracuse University. NY. Syracuse University Press (1967)

(4.) Coupe, RR, Hsu, B, "Penetration of Varnishes and Inks into Paper Under Pressure." JOCCA, 43 (10) 720-736 (1960)

(5.) Hsu, B, "Physical Surface Properties of Paper and their Relevance to Printablity." JOCCA. 46 (4) 263-275 (1963)

(6.) Oliver, JF, Forsyth, RC, "A Dynamic Liquid Sorption Apparatus for Studying Interactions of Microscopic Drops In Situ on Porous Substrates." Colloids Surf.. 43 295-305 (1989)

(7.) Clarke, A, Blake, TD, Carruthers. K, Woodward. A, "Spreading and Imbibition of Liquid Droplets on Porous Surfaces." Langmuir, 18 2980-2984 (2002)

(8.) Bikerman, JJ. "A Method of Measuring Contact Angles." Ind. Eng. Chem. (Anal. Ed.), 13 (6) 443-444 (1941)

(9.) Carre, A, Schreiber, HP, "Dependence of Surface Properties in PMMA Coatings on Preparation History." J. Coat Technol., 54 694 (1982)

(10.) Hutchinson, GH, "Prospects and Problems in Printing Ink Technology." JOCCA, 72 (7) 265-272 (1989)

(11.) Rossell, JB, "Factors Affecting the Quality of Frying Oils." In: Kochhar SP (ed) Conference of SCI Oils and Fats Group New Developments in Industrial Frying. Chapter 2. pp. 35-71. P.J. Barnes & Associates, Somerset. UK (1997)

(12.) Gil, B, Yoon, SH, Choi, ES, "A Fryer Design with Constant Specific Surface Area." Am. Oil Chem. Soc., 79 (5) 511-513 (2002)

(13.) Cannegieter, D, "Isomerization and trans-esterification." Paint Oil Chem. Rev., 110 (4) 17 (1947)

(14.) Hutchinson, GH, "Some Applications of Viscometry to Drying Oil Research." JOCCA, 42 (4) 283-301 (1959)

(15.) Davison, JW, "Flow Plate Viscometer 2007." Davison Chemographics Test Equipment, www.davisonchemographicsltd.com/more_testequipt.htm#flow

G. H. Hutchinson([??])

19 Broadlands Avenue, Southbourne, Dorset BH64 4Q, UK

e-mail: WallisSciences@aol.com

J. W. Davison

Davison Chemographics, Unit 28, Borden, Hampshire

GU35 9QF, UK

e-mail: John@davchemo.demon.co.uk