Viscosity Correlation for Aqueous Polyvinylpyrrolidone
`(PVP) Solutions
`
`Jason Swei, Jan B. Talbot
`Chemical Engineering Program, University of California, San Diego, 9500 Gilman Drive,
`La Jolla, California 92093-0411
`
`Received 14 May 2001; accepted 11 June 2002
`
`ABSTRACT: Polyvinylpyrrolidone (PVP) is characterized
`by its K-value, which is a function of the average molecular
`weight, the degree of polymerization, and the intrinsic vis-
`cosity. The viscosity was measured for aqueous PVP solu-
`tions with K-values ranging from 92.1 to 95.4 and with
`concentrations from 2 to 3 weight percent. A correlation was
`determined that relates solution viscosity to the K value and
`
`weight percent PVP, which is particularly useful in its use as
`a photoresist in the manufacture of high resolution display
`screens. © 2003 Wiley Periodicals, Inc. J Appl Polym Sci 90:
`1153–1155, 2003
`
`Key words: polyvinylpyrrolidone (PVP); viscosity; photore-
`sist
`
`INTRODUCTION
`
`(C6H9NO)n,
`Polyvinylpyrrolidone (PVP) polymers,
`are used by a wide variety of industries, such as
`pharmaceutical, cosmetic, textile, adhesive, coating,
`and ceramic. This is due to the unique physical and
`chemical properties of PVP, particularly its good sol-
`ubility in water and organic solvents, its chemical
`stability, its strong complexing ability with both hy-
`drophobic and hydrophilic substances, and its non-
`toxic character.1 PVP has been used as a photoresist
`system being activated with 4,4⬘-diazidostilbene-2,2⬘-
`disulfonic acid disodium salt (DAS).2 It has been uti-
`lized to produce higher resolution cathode ray tube
`(CRT) displays than the traditional photoresist, poly-
`vinyl alcohol (PVA) activated with ammonium di-
`chromate (ADC).3,4 The PVP/DAS system has the ad-
`ditional benefit of being more environmentally accept-
`able, since it is free of chromium and other toxic
`metals. However, not much is known about PVP prop-
`erties for this application, unlike the PVA/ADC pho-
`toresist, which has been studied for many decades.
`PVP is characterized by the K-value, or
`the
`Fikentscher’s viscosity coefficient, which is used
`mostly for polyvinylpyrrolidone and vinylpyrroli-
`done copolymers.5 The K-value is based on kinematic
`viscosity measurements, given by the Fikentscher
`equation6:
`
`Correspondence to: J. Talbot (jtalbot@ucsd.edu).
`Contract grant sponsors: Sony Electronics, Inc., and the
`University of California UC-SMART program.
`
`Journal of Applied Polymer Science, Vol. 90, 1153–1155 (2003)
`© 2003 Wiley Periodicals, Inc.
`
`log(crel)
`c
`
`⫽
`
`2
`75K0
`1 ⫹ 1.5K0c
`
`⫹ K0
`
`(1)
`
`where c represents the concentration in g/100 mL; ␩
`rel
`represents the relative viscosity as compared to the
`solvent; and K0 represents K/1000. The K-value can be
`directly calculated by rearranging the Fikentscher
`equation:
`
`K ⫽
`冑300clog(␩rel) ⫹ (c ⫹ 1.5clog(␩rel))2 ⫹ 1.5clog(␩rel) ⫺ c
`0.15c ⫹ 0.003c2
`
`(2)
`
`However, the determination of the K-value depends
`upon measurement of the viscosity in a specific con-
`centration with a particular viscometer.7 The concen-
`tration of anhydrous PVP is specified for each grade of
`the polymer: 0.1%, 1%, and 5% (w/v) for K⬎95,
`18⬍K⬍95, and K⬍18, respectively.
`The viscosity-average molecular weight, Mv, can be
`calculated from the empirical equation3:
`
`Mv ⫽ 22.22共K ⫹ 0.0075K 2兲1.65
`
`(3)
`
`These equations show that the higher the K-value, the
`higher the viscosity and the molecular weight. A PVP
`solution with a K-value between 80–100 has a molec-
`ular weight in the range of 900,000–1,500,000. In aque-
`ous solutions, the effect of PVP concentration on the
`viscosity shows nearly a first-order relationship with a
`much greater influence at large K-values.5,7
`The PVP/DAS photoresist is a dilute aqueous solu-
`tion of a high K-value PVP polymer and DAS activa-
`
`ARGENTUM PHARM. 1059
`
`000001
`
`

`
`1154
`
`SWEI AND TALBOT
`
`Figure 1 Viscosity versus weight percent PVP at room temperature (⬃23°C).
`
`tor. Small amounts of surfactants are also sometimes
`added. PVP is the component of the photoresist that
`significantly contributes to the viscosity of the pho-
`toresist solution. The photoresist is applied to a dis-
`play screen by spin coating and hence, the viscosity of
`the photoresist solution and the rotational rate of the
`panel control the thickness of the photoresist on the
`screen.3,8 Therefore, in order to get a consistent pho-
`toresist thickness, the viscosity of the PVP solution
`must be known and kept constant.
`PVP obtained from a vendor with a nominal K-
`value actually has a wide range of K-values that fall
`within their specifications. For example, BASF sells
`PVP with a nominal K-value of 90, but the actual
`K-value can range anywhere between 90 and 98. Since
`the viscosity, weight percent of PVP, and K-value of a
`PVP solution are related, with each differing K-values,
`the weight percent of PVP in the photoresist solutions
`must be raised or lowered to compensate. In order to
`consistently control the viscosity of the PVP polymer
`solution more accurately, a correlation was required to
`relate the K-value, viscosity, and weight percent of
`PVP.
`
`EXPERIMENTAL
`The PVP solutions were obtained from BASF as pre-
`mixed solutions with about 20 wt % PVP solids dis-
`solved in water. Four different K-values of PVP were
`used: 92.1 of an unpreserved PVP, 92.5, 93.6, and 95.4
`of a preserved PVP. These K-values of PVP represent
`the range that is currently of interest for its use as a
`photoresist material. The preserved PVP solutions
`contained the biocide Cosmocil, which is used to in-
`crease shelf life. These PVP solutions were then fur-
`ther diluted with de-ionized water to produce solu-
`tions between 2 and 3 wt % PVP. First, the appropriate
`
`amounts of original PVP solution and water were
`determined to produce the desired concentration of
`PVP in solution. A 100 mL beaker was placed on top
`of a balance and the desired amount of water was
`weighed, followed by the addition of the determined
`mass of PVP. This solution was then mixed with a
`magnetic stir bar until homogeneous (about 30 min).
`A Brookfield Syncro-Lectric viscometer, model
`#LVT, with an ultra-low (UL) adapter was used to
`measure the viscosity of the solutions. To measure the
`viscosity, 16.5 mL of solution was placed into the
`capped UL adapter. The UL adapter was then placed
`around the spindle and attached to the viscometer.
`Once assembled, the viscometer was operated at the
`three highest possible spindle rotation rates (3, 6, 12,
`and 30 rpm depending on the viscosity of the solution)
`to increase the accuracy of the reading, and the vis-
`cosity was measured at each rate. The measurements
`at these three rates were then averaged to determine
`the viscosity of the solution. The accuracy of the mea-
`surements was ⫾0.5 cP.
`
`RESULTS AND DISCUSSION
`The viscosity was measured at varying concentrations
`of PVP between 2 and 3 wt % at room temperature
`(⬃23°C). Figure 1 shows that the viscosity has a sec-
`ond-order polynomial dependence upon the weight
`percent of PVP for all K-values. The correlation for
`each K-value of PVP is listed in Table I. Figure 1 also
`shows that the viscosity increases for each weight
`percent of PVP as the K-value is increased.
`With this data, a correlation relating the K-value,
`viscosity, and weight percent of a PVP solution was
`determined. The data were processed using a statisti-
`cal analysis package (DOE KISS), and the following
`
`000002
`
`

`
`VISCOSITY CORRELATION FOR PVP SOLUTIONS
`
`1155
`
`correlation was determined (with the coefficients
`rounded off to four significant figures):
`
`␮ ⫽ 38730W 2 ⫺ 9551W ⫹ 95.51WK ⫺ 32.17K
`⫹ 0.1641K 2 ⫹ 1581
`
`(4)
`
`where: ␮is viscosity (cP), and W is the weight percent
`of PVP. Although this correlation is written to deter-
`mine viscosity, it can be easily rearranged to deter-
`mine the K-value. This equation is a second order
`polynomial in K-value and can be solved by using the
`quadratic equation as follows:
`
`共0.1641兲K2 ⫹ 共95.51W ⫺ 32.17兲K
`⫹ 共38730W 2 ⫺ 9551W ⫹ 1581 ⫺ ␮兲 ⫽ 0
`⫺ b ⫹ 冑b2 ⫺ 4ac
`2a
`
`K ⫽
`
`(5)
`
`(6)
`
`where a ⫽ 0.1643, b ⫽ 95.51W ⫺ 32.17, and c
`⫽ 38730W2 ⫺ 9551W ⫹ 1581 ⫺ ␮. The validity of this
`correlation was determined by measuring the viscos-
`ity of other PVP solutions with known K-values and
`weight percents of PVP. The data from these experi-
`ments are summarized in Table II.
`The viscosity was calculated by using the correla-
`tion in eq. (3) for three different weight percents of
`PVP and then compared with the average measured
`viscosity. The average relative error for this correla-
`tion was approximately 1%. The viscosity was also
`calculated using the Fikentscher equation , which gave
`viscosities with an average relative error of about 58%.
`Equation (3) yields a much better prediction of the
`viscosity for the narrow range of conditions studied,
`which are of use for the photoresist application. How-
`ever, by checking the predictions of both of these
`correlations with data available in the literature,5,7
`neither was found to be a general correlation. There-
`fore, a correlation, such as given in eq. (3), would need
`
`TABLE I
`Correlations of Viscosity as a Function of
`Weight Percent PVP
`Correlation
`
`␮ ⫽ 45300W2 ⫺ 1000W ⫹ 14.2
`␮ ⫽ 30200W2 ⫺ 187W ⫹ 3.85
`␮ ⫽ 51500W2 ⫺ 1080W ⫹ 15.0
`␮ ⫽ 28000W2 ⫺ 248W ⫹ 5.66
`
`K-Value
`
`92.5
`93.6
`95.4
`92.1(U)
`
`R2
`
`0.9996
`0.9988
`0.9988
`0.9999
`
`TABLE II
`Viscosity Correlation Data
`PVP
`Average
`Calculated
`Viscositya
`Concentration
`Viscosity
`(cP)
`(Wt %)
`(cP)
`
`Fikentscher
`Viscosityb
`(cP)
`
`2.22
`2.65
`2.93
`2.13
`2.63
`2.88
`2.18
`2.62
`2.95
`2.15
`2.60
`2.97
`
`14.4
`19.1
`23.4
`13.6
`19.5
`23.5
`15.9
`22.0
`27.8
`13.3
`18.1
`23.6
`
`14.2
`19.3
`23.3
`13.9
`20.0
`23.8
`16.1
`22.4
`28.0
`13.5
`18.4
`23.6
`
`19.7
`30.8
`40.5
`18.9
`32.2
`41.4
`22.0
`35.5
`49.8
`17.9
`28.6
`41.0
`
`K-Value
`
`92.5
`
`93.6
`
`95.4
`
`92.1 (U)
`
`a Calculated from eq. (3).
`b Calculated from eq. (1).
`
`to be developed for the particular range of conditions
`of interest.
`
`CONCLUSIONS
`A correlation between the viscosity, K-value of the
`PVP, and weight percent of the PVP was determined.
`Since an unconventional method of characterization,
`the K-value, is used for the PVP polymer, the K-value
`was related to more standard and convenient charac-
`terization method of viscosity measurement. With the
`K-value of the PVP, as given by a vendor, the corre-
`lation can be used to calculate the appropriate concen-
`tration of PVP needed to produce the desired viscos-
`ity, and thus allowing for the deposition of a uniform
`coating thickness of the photoresist.
`
`References
`
`1. Kroschwitz, J. I.; Howe-Grant, M., Eds. Kirk-Othmer Encyclo-
`pedia of Chemical Technology, 4th Ed.; Wiley: New York, 1989,
`Vol. 24.
`2. Akagi, M.; Nonogaki, S.; Koshi, T.; Oba, Y.; Oikawa, M. Polym
`Eng Sci 1977, 17, 353.
`3. Castellano, J. A. Handbook of Display Technology; Academic
`Press, Inc.: San Diego, 1992.
`4. Ozawa, L. Cathodoluminescence: Theory and Applications;
`VCH Publishers: Weinheim, Germany, 1990.
`5. Kroschwitz, J. I., Ed. Encyclopedia of Polymer Science and
`Engineering; Wiley: New York, 1989, Vol. 17.
`6. Kline, G. M. Modern Plastics 1945, Nov., 157.
`7. Fusiak, F., Ed. PVP; In Polyvinlypyrrolidone Polymers; Interna-
`tional Specialty Products: Wayne, NJ, 1999.
`8. Castellano, J. A. Handbook of Display Technology; Academic
`Press, Inc.: San Diego, 1992.
`9. Sasaki, K. Y.; Talbot, J. B. Adv Mater 1999, 11, 91.
`
`000003