You are on page 1of 4

J. Chem. Eng.

Data 1909, 3 4 , 371-374

371

Complex Permittivity of Water as a Function of Frequency and
Temperature
Udo Kaatze
Drittes Physikalisches Znstitut, Universitat Giittingen, Burgerstrasse 42-44, 0 3 4 0 0 Giittingen, West Germany

Sets of complex permittlvlty values for pure water are
presented for nlne temperatures between 0 and 50 O C
and for mlcrowave frequencles from 1.1 to 57 GHz. The
data are analyzed together wlth static permittlvlty values
and many sporadlc mlcrowave data found In the published
literature. At all temperatures considered In this analysls
(-4.1 OC IT I 60 "C) the frequency-dependent complex
permittlvltles can be well descrlbed by a Debye-type
relaxation spectral functlon reflectlng one discrete
relaxation tlme. The parameter values for this functlon as
obtalned by a nonllnear regression analysls of the
measured dlelectrlc spectra are displayed In tabular
format. Also given are emplrlcal correlations allowing
these parameters to be Interpolated wlth respect to the
temperature.

I ntroductlon
The insight into the molecular properties of water has a
substantial influence on ecological, industrial, and medical developments. This unique liquid therefore plays a dominant role
in many fields of scientific research. Our increasing interest,
of course, is not only dlrected toward the unsolved mysteries
of pure water itself. Considerable effort is undertaken to elucidate the molecular structure and dynamics of solutions, both
electrolytic and biomolecular. A fundamental precondition for
many studies of aqueous systems is the precise knowledge of
the pure solvent properties. Water is also a most suited reference liquid in calibration and test measurement.
The complex (electric) permittivity

= E ' ( V , T ) - it"(v,T)
(1)
measured as a function of frequency v and temperature T is
E(V,T)

of interest for various reasons. I t shows how the liquid responds to electromagnetic waves, one of our most important
means of communications. In addition, the microwave part of
the dielectric spectrum presents distinct dispersion (dt'(v, T)/dv
< 0) and absorption (t"(v,T) > 0) characteristics. These are
due to restricted orientational motions of water molecules.
Measurements of the permittivity of aqueous systems as a
function of frequency and temperature enable conclusions on
relevant molecular interactions and arrangements.
Because of the importance of the permittivity of water in
many fields of interest, the available data had been collected
and critically evaluated some years ago (7). That review,
however, aimed at the correct analytical description of the
water spectrum by a physically meaningful relaxation function
rather than a compilation of experimental results. I n various
applications there is an additional demand for original data.
These are given in this article. Besides the previous results
from this laboratory, recently measured Permittivity data at
further temperatures are also displayed. Empirical correlations
are presented which allow for the interpolation with respect to
both frequency and temperature. Literature data are referenced and Included in the analysis.
Experlmental Methods
The water was bidistilled and deionized by bed ion exchange.

Its specific electric dc conductivity u was smaller than 2 X lo4
0021-9568/89/1734-0371$01.50/0

S/m. Contributions u/(tow)to the t" values are thus negligibly
A s V-' m-'; w = 2x24.
small (to = 8.854 X
Spot frequency measurements have been performed by
using a traveling-wave technique as first proposed by Buchanan
(2) and later on improved by Pottel(3). As schematically shown
by Figure 1, a propagating electromagnetic wave transmitted
through an appropriate specimen cell of variable thickness was
probed interferometrically. We used cells (6 in Figure 1) that
essentiaily consist of a piece of circular cylindrical waveguide
or coaxial line. I t was sealed agalnst the feeding line by a
window (7) made of Teflon, silicone rubber, or ceramic dlelectrics. A receiving probe (8), also a circular waveguide or coaxial line, was immersed in the liquid. Provided with high-precision ball bush guides this probe was shiftable free of backlash
along the direction z of wave propagation.
At a suitable starting position z = s, of the probe the reference wave of the double-beam interferometer was adjusted
so that the probe signal and the reference signal interfered to
zero. The reference branch contained a phase shifter (9) and
a microwave attenuator (IO) for this purpose. I t was addltionally provided with a calibrated attenuator (1I), the phase shift
of which was constant. When the probe was moved for the
wavelength A and the calibrated attenuator was adjusted for the
attenuation per wavelength, ax,zero signal was again obtained.
The Permittivity of the sample liquid is related to the measured quantities v, A, and aA according to the equations

t'(v) =

[ ;I2 2[I:
-

+

[];

2

(2)

aAt
t"(v) =

7rA

(3)

where A, = (speed of light)/v denotes the wavelength of free
space and X, the cutoff wavelength of the empty cell.
Seven double-beam interferometers, each matched to a Ilmited frequency range were used. One was constructed with
coaxial line components, and the others were constructed with
standard waveguide devices. For reasons of comparison at
least two different cells and two different calibrated attenuators
were available for each interferometer. All cells were surrounded by a coat for circulating thermostat fluid to enable the
temperature of the sample to be controlled to within f0.05 K.
Uncertalnty of Prlmary Data
Preckrlon. High stability in both the mechanical and electronic components of the interferometers together with a sufficiently effective temperature control and a very sensitive superheterodyne zero signal detection allowed for an adequate
reproducibilityin the measurements at all frequencies bebw 26
GHz. An example for the precision of repeatedly measured
data is shown in Figure 2. The spread in the values A, (n =
1, ..., 17) for the wavelength within the liquid is smaller than
f0.1% in this case. I n the values (ax),of the attenuation per
wavelength the spread is smaller than 0.3 % . Due to the higher
aA values at concurrently reduced receiver sensitlvlty, the
measurements were less reproducible above 26 GHr. At 40
GHz, for Instance, the scatter is f2% in both the A, and (a&
values.

0 1989 American Chemical Society

values should characteristically depend on the starting positions s. Schematic representation of the microwave doubbbeam interferometers. 14. Experimental Uncertainties of the Complex Permittivity Values for Pure Water. Likewise precisely known was the position z of the probe within the liquid. and (ah).3-12.. (b) contributions from multiply reflected signals.3 5. As has been realized previously (45). respectively. 1989 1 3 9 10 11 13 Figure 1. Table I.crcalmj(vn ))I + Nft . I f one of the three fist effects acts as a noticeable influence on the measurements. 10. 3. err. Remaining sources of possible systematic errors were (a) small deviations from completely coaxial movements of the probe.. No. I t was found that the dielectric spectra can be sufficiently well represented by a Debye-type relaxation function.-s~)/A I I I] 12 14 16 Flguro 2. Resutts and Semlempirlcal Representatlon of Data I n Table I1 the primary complex permittivity data from this laboratory are compiled as a function of frequency and temperature. Between these runs either the specimen cells or the calibrated attenuators of the microwave interferometers have been exchanged. frequency counter or frequency meter. variable attenuator. 4. No systematic errors exceeding the limits of reproducibility of the data have been found thereby. Single values at 6. In addition to this test. and negative imaginary part. 4. 1.5 4 10 3 1 1 1 2 3 4 with different interferometersare matched. and (d) imperfections of the calibrated attentuators either in the reading of the adjusted attenuation or in the invariability of the phase shift. monochromatlc oscillator. uniline or coaxial attenuator.. the errors given in Table I were found. (c) the presence of higher order fieid modes within the sample.this is also true if far-infrared data are included in the analysis. (n = 1. d. differently constructed spechnen cells have been used in each frequency band to look for possible influences from sources a-c. As demonstrated by Figure 2. 12.1-5. GHz Ae'lc'. dielectric window. (ax).. Also shown are the mean values and ax.. for water at 15 OC plotted as a function of the starting position s. 44) permittivity values. sample cell with 7. phase shifter.l n = l 1 Ne. v= C [ Ws'(vn ) ( ~ ' m a a ( v n ) . 9. 13. .=1 (5) . As shown by Figure 3. calibrated variable attenuator with constant phase shift. and 8. superheterodynereceiver. directional coupler or coaxial beam splitter.372 Journal of Chemical and Englneerlng Data.1. As predicted by eq 4. Equation 4 has been fitted to the sets of measured permittivity values by a nonlinear least-squares regression analysis causing the variance 1 Ne.5 12. C [ ~ c " ( ~ n ) ( ~ ~ ~ m e -a ~crrcala(vn))I ( v n ) Nfu . of the complex permittivity displayed as a functlon of frequency v for water at 25 O C . The frequency was determined and kept constant during the measurements with a negligibly small error. the A. this has been done in great detail at 25 O C .flexible waveguide or coaxial line. Different figure symbols indicate different runs of measurements. of the probe. Real part. As a result of this analysis.5-18 18-26 26-30 30-40 40-57 [ Li 1 I 2 2. 9-42) and static (43. The data have been analyzed together with previously publishedmicrowave (2. 34. Accuracy. Imperfections of the calibrated attenuators have been considered by using different devices (up to four) in each frequency band and also by analyzing how well adjoining data measured J I 40GHz7O V Figure 3.457 GHz of the wavelength A. zero signal indicator. 6.5 1. precisely shiftable probe. this was obviously not the case. Also shown by bars is the effective frequency band for each interferometer. Vol. The Debye function (46) is given by the relation (4) To show that the measured spectra are compatible with this function. the permittivity values at microwave frequencies for water at 20 OC are displayed in an d(v) versus v d r ( v )plot in Figure 4. % AP/P.. as well as bars indicating the range of scatter in the data. % 1 1 1. impedance matching transformer. 5. 17)of the probe. 2. Effects of Possible Systematic Errors Included Y. and attenuation per wavelength. 11. 20 10 56 10 5 4 1% i 0 2 1 1 1 4 7 10 20 I 02 I 04 06 1 1 08 10 (s. a straight line is defined by the data.

424 6.9 31.44 39.9 34.0 72.08 38. We4vn) = )/At'(% 1 (6) N8 c )/A+.681 7.78 41.6 36.9 16.6 48.5 26.0 77.2 34.97 37.5 37.2 28.50 62.6 24. The dependence of these parameters upon the temperature T can be empirically described by the expressions €(o) = )/AE"(h1 NP (7) C t"-d(Yn)/At"(Vn) n=l using the experimental errors given in Table I or quoted by the €(OD) 7 ~~(l.5 59.49 7.43 15.7 34.14 27.0 19.73 22.004 6.5 76.6 21.59 26.2 28.9 77.5 69.61 28.57 8. The values obtained by this fitting procedure for the low-frequency permittivity t ( O ) .1 33.6 20.94404-l.1 50.6 17.414 6.8 31.7 73.4 14.38 25.5 36.5 17.6 71.5 25 O C (1) 21.9 38.84 36.2 19.9 3.6 26.8 19.850 6.7 35.1 25.48 58.79 26.305 66.7 67.3 32.4 21.6 11.3 30.8 26.3 51.318 7.9 29.889 8.628 2.9 22.950 8.6 55.6 70.0 31.7 12.63 26.638 5.4 53.843 5.7 16.47 17.376 5.17 28. 1989 373 Table 11.800 2. GHz t'(v) P(v) Y.000 6.00 22.9 36.850 7.8 78.77 9.707 12.1 39.2 33.5 48.43 26.750 3.1 72.7 71.2 70.2 24. N f land N.4 75.65 K)*) exp((2.2 36.738 8.44 56.0 26.02 12.9 28.56 36.5 52.2 26.53 26.41 19.0 34.62 52.1 41.800 3.9 74.1 34.2 34.32 71.46 56.0 29.1 76.1 1.9 77.3 33.) have been calculated according to the relations t'meard(Y.0 67.786 85.406 7.82 15.3 71.77 25.536 5.45 11.53 24.8 58.5 34.5 45.200 2.56 42.610 2.39 17.7 24.8 18.4 72.8 7.6 34.6 31.214 79.8 67.04 16.3 11.9 38.5 35.565 v.7 27.23 25.3 27.6 17.2 16.7 32.2 18.7 80.176 2.9 70.5 38.3 70.462 79.3 3.3 31.3 13.7 73.97 39.9 28.508 10.1 34.0 53.8 1. denote the number of E' and E" values.7 14.8 67.84 37.3 12.8 6.273.67 26.5 20.340 7.4 53. GHz d(v) c"(v) 0 "C ( 4 ) 7.1 35.1 33.2 61.2957 X lo3 K)/T) (10) .6 47.901 5.3 35.1 59.55 15.733 3.1 83.457 79.5 38. Complex Permittivity Data for Water as a Function of Frequency and Temperature Y.22 9.7 35. respectively. the extrapolated high-frequency permittivity E( a).433 5.3 15.3 12.7 15.0 12.8 34.64 26.9 30.146 3.0 49.85 39.7 7.5 57.853 6.0 17.8 34.1 23.8 35.83 27.4 45.31 26.7 34.594 1.660 10.7 12.47 21.828 75.1 66.6 76.6 12.8 67.5 13.6 50.3 21.78 28.9 13.6 56.5 29.6 34.4 20.8 37.2 23.900 7.4 29.729 6.703 16.7 69.53 24.4 42. GHz e'(v) ~"(v) v. 4.9 33.7 17.4 17.47 67.02 31.3 33.0 52.688 53.516 10.6 40.9 67.24 25.77 .25 11.7 68.8 31.8 29.3 34.2 21.75 28.8 67.0 36.1 36.3 74.78 44.7 15.70 26.1 58.1 35.723 5.54 17.00 15.368 6.4 21.8 12.4 47.83 36.81 37.1 6.345 7.1 37.1 20.82 25.922 5.49 37.7 32.579 8.7 38.0 61.515 3.300 6.0 30.5 34.8 31.4 63.2 7.7 69.1 12.2 47.7 14.0 45.) ~'Inbad(% n=1 and E"mgasd(% Wf.9 14.65 36.150 1.8 34. Vol.7 72.7 18.4 16.3 69.6 19.3 75.5 10.2 36.2 64.9 5 " C (5) 1.5 13.2 68.2 37.8 79.1 36.6 37.58 36.9 7.38 30.2 23.6 56.8 29.3 3.41 9.7 33.2 27.6 28.2 17.300.7 19.5 25. Also included in this table are some data at temperatures at which we have published experimental permittivity values only ( 7 ) .1 30.9 33.1 53.0 11.13 28.5 36.5 6.59 13. The weighing factors We+.06 17.8 65.6 34.36 65.02 21.15 K)) = 5.3 35.64 69.9 22.000 2.941 7.3 54.6 36.39 8.77 24.40 23.0 29.4 33.7 36.345 6.60 27.914 2.412 78.9 33.9 7.0 16.45 26.745 x (8) (9) + 10-15 s)(l (7 x 10-5 K-2) x ( T .6 67.25 57.0 77.f(V") = authors of literature data.5 28.80 24.306 5.7 15.94 17.8 32.8 62.80 36.6 12.32 10.0 34.979 9.0 21.01 23.5 21.74 X lo-* K-l(T.2 35.5 3.9 79.1 7.1 73.8 37.8 68.5 26.23 41.23 10.2. No.3 68.1 73.958 7.2 12.4 34.3 19.1 33.56 13.8 26.14 25.3 34.8 20.8 75.3 12.4 14.267 7.6 29.6 15.8 55.5 14. respectively.1 30.8 35.145 78.6 75.2 56.991X10JK"(T-273.8 15.3 18.7 35.9 17.73 25.5 3.244 8.4 33. GHz d(v) c"(v) 26.738 71.4 76.and the relaxation time 7 are collected in Table 111.2 58.4 36.6 12.4 33.8 31.01 48.6 72.Journal of Chemical and Engineering Data.300 5.922 5.30 34.9 20 "C (6) 13.2 32.3 18.4 32.15 16.6 31. 34.86 10.137 3.15 K) = (3.1 11.15 28.4 14..5 58.1 32.43 30.1 5.6 13.4 19.49 23.762 76.7 5.56 33.497 6.6 10 " C 15 O C 30 "C 35 O C (7) 50 "C (8) to adopt its absolute minimum.9 71.) and W&.50 23.2 26.6 21.780 13.151 3.6 44.821 2.35 38.8 20.6 37.7 17.2 19.2 38.417 3.45 60.8 8.4 32.0 69.1 13.9 34.623 3.

R. Kaatze. Res. I n this paper the high-frequency permittivity is just defined to allow for the interpolation of the permmMty values below 100 GHz. C. Y.. Final Report to the BCR Project 148 Community Bureau of Reference: Brussels. (Lelpzb) 1965. D. 133.2 0 87. Chem.. U. 1980. 1988. Phys. Chem.05 8. Losen. Straiton. Ser.3 4. 726.. Moreau. 967. 1971. Cdlle. R. 297. Uhlendorf. J.Phys. 156. P. W. E. Ser. Phys. J. Unlversity of W n g e n . El-Sabeh. Appl. Bottreau. J...3 5.2 5. Phys. Br.27 f 0. 1957. if measured data of the near-millimeter-wavelength region are Induded in the analysis (45). B 1973. S. C. J. 151.... Polar Mo/ecules. Chem.83 f 0. J.21 0. Bottreau.1 15 82. (h4unM) 1981. J. J. 73. B 1953.V. A.-M. Appl. Oottlob. Grant. Blue.7 f 0.2 4. 273.H.1 89. 1792. Barthel. Appl. Community Bureau of Reference: Brussels. S. 6 5 . 6 5 . Botbeau. 1691. Llttle. P. E. Van Loon. D.. C. Farady Soc. 1973.69. Vol. Hoffmann. 85. Appl. 76. H. Kaatze. Welss.4 7. 1929. A.2 f 0. 1974. R. U.3 6. Phys. F. 249. 1 7 . G. J. Lane. H. Owen. University of W n g e n .05 f 0. R. Flnsy. 1965. Moreau.1 10 83. D. 291.. 4 4 . 4 9 . Chem. 7 7 .. Phys. Sexton. Marzat. Accepted June 8. W. Inst. e'. Marzat.. Sheppard. Vldullch. C.87 f 0. F. Dlpbm-Thesls. 881. Ser. S.T) within the present limits of experimental errors. 6 4 ..0 f 0. (Munich) 1970. Buchanan. H.O C c(0) f As(0) e(=) f As(=) T f AT.. G. Smyth.F. R. Van Loon.50 f 0. J. H. A. A 1952.0 f 0. 0. Meteorokghl Factors In Redk-Wave Ropapilon. J. E. Grant. Grant. 60.. J.2 3. Sheppard. 1953. H. Chem. 8.2 4. Chem.. Phys.. Trans. 276. J.7 f 0. Grant. 1961. B r . Appl.70 f 0. Scl. J . Phys. Instrum. D. M. B. Final Report to the BCR Project 148. 2 . Roc. H. Soc.2 6.18 f 0. 241.05 70 \ 60 - -L I 5 50 - 40 - 30 - "\ *\ y* 20 10 with previously published experimental data. Smyth. Buchanan. 175. Electr. Gerdes. F. 1974.. 6. 132. P. 1973. Schwan. Rltson.5 10. Grant. cook. Kay. Relation 8 was originally derived for static permittivity values measured between 500 Hz and 100 kHz (44). 956. The e x p " W data have been taken from the literature (e. Chem. Debye. I . Kaatze. A. Sci. The Physical Soclety: London.2 14. Am. where the predictions of the empirical functions are compared Kaatze. Chem.2 5.10 f 0. Phys.. Unlversity of Gatthrgen. Dlpbm-Thesis. J. 74. Y.. 9 . E. 1980. Hasted. J. K.M. Roc. H.2 5. of the complex permmMty of water at 9. 363. Scl. Chemlcal Catalog: New York. 163. J. H. Phys. D 1972. 71. Dlplom-Thesls. I Literature Cited 70 40 65 35 60 c. J. 65. Hallenga.375 OHZ shown as a fundon of temperatwe T.9 f 0. M. Sci. 6 6 . A.. Cook. No. R.82 f 0. The dm) values inferred from the microwave data are less reliable with respect to the far-Infrared properties of water. 5 . 1975.374 Journal of Chemical and Engineering Data.1 78. H. P. S. The MI curves represent the predictions of the empirical relations 8-10 and 4. Lett. 1163. C. Recehred for revlew November 29.-M. Soc. K.. for instance. Water. Lacroix.. Bellemans. Y. R. P. Ph... NigMlngale.Lubitr.05 35 74. Roc. 1576.London. Dutult. Appl. 1976. 2257. P. Phys. 70. 0. Conner. B.. 1807. and negative imaslnary part. J. L. C. C. 1204. 2. 1967. J..25 f 0. J. W. E. Part 3 1952. J. Srwarnowskl. Chem. unhrerslty of Qbttlngen. Sheppard. Bottreau.1 and 60 OC relations 8-10 together with eq 4 can be used to calculate the complex permmMty e(u. J. A. Szwarnowskl. Saxton. J . Z . Chem. Chem.2 12. Lacrolx. 30 55 E'' 25 50 20 0 10 20 30 40 50 "C 60 T Figure 5. 70. F. for thls relaxation function corresponds to an exponential dielectric decay function that cannot correctly account for the short-time behavior of the electric polarization. R. 4. Cogan. T. Phys. 1952. Phys. 656. London 1948. Mllner. London.01 f 0. 1984. N. K. Marzat.. B 195& 4 . 81. H. R. The high-frequency permittivity values. 1959. Tolbert. J. Microwave Power 1978. 2 6 . E. 1969. Ber. Delbos.. DS -4.. J. M. 255. 1101. Marzat.56 f 0.1 f 0. 4 6 . 1975. V. 34. Sandus. Rev. Phys. Vial. E. 1959. Phys. A.36 f 0. Kestemont. 7732-18-5. 1003. 73. Glese.91 f 0.3 21.6 f 0. At frequencies below 100 GHz and temperatures between -4. 9S. Chem. J. Evans. A. E 1980. Rampolla.. W. Microwave Power 1975. 70.89 f 0. E. C..2 5. R. Hertel.. 145. Phys..2 20 80. 7989 90 7 Table 111. 0. Thhe de trolsieme cycle. 1989.9 f 0. Sheppard. C. Miller. A.05 50 69. Miller. Soc. Poky.2 5. Soc.91 f 0. 115.. Miiller. 51. J. J . Finsy. V. 1946. R. J. R. An example is given in Figure 5..28 f 0.68 f 0. 2 4 . U. the use of the Debye function (eq 4) in the extrapolation to high frequencies has to be questloned. 6 0 .. J. Grant. 1953.2 9. Shack. 1. Gmphys. C. 2085. Univeralty of W n g e n . Dlplom-Thesis. G. C. e". Appl. Diptom-Thesls. . J. 30. J. 566. Instrum. H.. J. Oiese. Phys. A.3 4. R. Appl. J. Phys. H.5 f 0. Ooulon. T. H. R. lB66. Delbos. Bunsen-Ges. Phys. E 1977.. C. 228. 382. Y. Eng.. 9. J.2 f 0.. J..05 40 73. A.2 f 0. E 1981. M. Roc. Such changes are indeed of minor importance for the predictions at microwave frequencies but might be relevant in the discusdon of the water spectrum above 100 GHz. 3 . Sheppard.-M. Chem. Behret. Real part. Lane. Dutult.ref 2. Rev. 1980. phvs.1 5 85.2 5. 400. Y. 1966.346 GHz (77)).Phys. L. Acad. Phys.R. R. Adolph.. R. 156. Dutult...2 5.J. Rev. Moreover. 1976. Schmlthals. Rev.2 17.83 f 0. J. 1987. change for a small amount.05 60 66.. 1989. 87.. 1943. 337.05 25 5. C. R. Phys.. Ser. Phys. Br.75 f 0. 1961. Unlverslty of Bordeaux. Pottel. Phys. 6 9 .. Hasted. E.67 f 0. - Regbtry No.. Res... Parameters of the D e b y e Relaxation Spectral Function (Eq4) for Water at Various Temperatures T. D. J. U.36 f 0. 6...92 f 0.02 30 76.