Journal of Magnetism and Magnetic Materials 209 (2000) 71}74

Magnetic water treatment

J.M.D. Coey*, Stephen Cass
Physics Department, Trinity College, Dublin 2, Ireland

Abstract

Carbonates formed by heating water containing +120 mg(Ca)/l are characterized by X-ray di!raction and electron
microscopy. Tests on 32 pairs of samples establish, at the 99.9% probability level, that drawing water through a static
magnetic "eld (B+0.1T, +B+10 T/m) increases the aragonite/calcite ratio in the deposit. There is an incubation period
of several hours, and memory of magnetic treatment extends beyond 200 h. ( 2000 Elsevier Science B.V. All rights
reserved.

Keywords: Magnetic water treatment; Superparamagnetism; Permanent magnet applications

The limescale problem in hard water arises because the CaCO , with rhombohedral and othorhombic crystal
3
solubility of CaCO decreases with increasing temper- structures, respectively. Aragonite has the higher density,
3
ature [1]. Huge amounts of energy are wasted because and it is less prone to form hard scale. The e$cacity of
hard scale forms in boilers, heat exchangers and domestic magnetic treatment is reported to last from tens of min-
hot-water systems. Various magnetic, electromagnetic utes to hundreds of hours. There is a review of the
and electrostatic devices purporting to control limescale literature by Baker and Judd [2].
formation are sold worldwide for domestic and industrial These, and similar claims of a "eld e!ect on precipita-
applications. Typical products incorporate arrangements tion of other salts, coagulation of colloids and wax
of permanent magnets; large magnet structures are in formation from crude oil, have been met with consider-
daily use in industries ranging from brewing to hydro- able scepticism, mainly because there is no obvious way
electric power generation. Belief in the bene"cial e!ects of for a magnetic "eld to in#uence any of these processes.
magnetic "elds on water has led to the sale of millions of Much of the irreproducibility of the data, and possibly
magnetic cups in China. the e!ect itself, may result from inadequate control of
Despite its ubiquity, there is relatively little scienti"c experimental conditions. Here, we set out to establish
literature on magnetic water treatment. It is not clear whether or not any e!ect exists. We conducted blind tests
how, or even if, it works. Unlike chemical water soften- using identically treated pairs of samples, with and with-
ing, magnetic treatment should have no direct e!ect on out magnetic "eld. There is considerable variability in the
water chemistry (unless the magnets are in contact with results, but our method allows us to answer the key
the water); yet, it is claimed to alter the morphology and question, and to identify some relevant variables.
adhesion of calcium carbonate scale. Published data are Two groups of experiments were conducted, each us-
often contradictory. For example, there is some dispute ing di!erent water and a di!erent magnetic device. The
as to whether the deposits of calcium carbonate from "rst was on groundwater drawn from a well sunk in
magnetically treated water are predominantly calcite or limestone in West County Dublin, Ireland. The water
aragonite. These are the two common natural forms of could either be drawn through a plastic "lter assembly
containing a stack of Te#on-coated ferrite ring magnets1,
or bypass the magnet assembly. The water was sealed in

* Corresponding author. Tel.: #353-1-6081470; fax: #353-

1-6772941.
E-mail address: jcoey@tcd.ie (J.M.D. Coey) 1 Manufactured by D-Fluid, Milan.

0304-8853/00/$ - see front matter ( 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 3 0 4 - 8 8 5 3 ( 9 9 ) 0 0 6 4 8 - 4
72 J.M.D. Coey, S. Cass / Journal of Magnetism and Magnetic Materials 209 (2000) 71}74

Table 1
Analyses of untreated water (mg/l). The only signi"cant changes after magnetic treatment are shown in square brackets

Well water Mineral water Well water Mineral water

Na 6 15 SO 15
3
Mg 24 16 NO 0.05
2
K 1 3 NO (0.01 9
3
Ca 132 114 Cl 23
Fe 2.49 [1.39] (0.01
Mn 0.46 [0.37] (0.01 pH 7.2 6.9
Zn 0.04 p (mS/m) 64.8

1-l polythene bottles, and all tests were conducted blind.

The experimenter (SC) was unaware whether a particular
sample had been drawn through the magnetic device,
or through the bypass valve. The second group of
experiments was conducted on a commercial still mineral
water, which was simply poured out of its 500 ml bottle
with or without a 20 mm split-ring collar containing
Nd}Fe}B magnets2 "tted around the neck of the bottle.
In each case the water was exposed to a maximum "eld of
+0.1 T, and a "eld gradient of +10 T/m. Analyses of
the water are given in Table 1. The only discernable Fig. 1. A set of data showing evolution of the aragonite ratio
A with incubation time t in untreated water (#), treated water
di!erence after magnetic treatment was a decrease in Fe *
(L) and the di!erence (d).
and Mn content of the well water, shown in square
brackets.
More than 100 samples were examined using a simple
protocol. Water was stored for a time t before heating in
*
open 500 ml beakers at 803 to form limescale. Scale was
collected on a fresh microscope slide at the bottom of the
beaker. All samples were examined by X-ray di!raction
using Cu K radiation, and 14 of them were selected for
a
scanning electron microscopy and microprobe analysis.
The ratio of calcite to aragonite was evaluated by
measuring the ratio of three X-ray re#ections in the
region 253(2h(303. The peaks are calcite 104 and
aragonite 111 and 102. The quantity

A"(I #I )/(I #I #I )
111 102 104 111 102
gives an estimate of the fraction of calcium carbonate Fig. 2. Electron micrographs of carbonate deposits from un-
treated (left) and magnetically treated (right) mineral water.
present as aragonite.
The incubation time t was varied from 0 to 200 h, and
*
the speed of #ow of water through the magnets was untreated (A"7%) and treated (A"54%) mineral
varied from 0.04 to 1.2 m/s. A ranged from 0% to 100%. water are shown in Fig. 2. The long, acicular crystals
No systematic in#uence of #owrate could be discerned, (+30]3]3 lm3) are identi"ed as aragonite [3], where-
but the data suggest a maturing e!ect when A increases as the equiaxed crystals (+4 lm) are calcite. Typical
with t for several hours, and that a signi"cant increase in
* microanalyses are shown in Table 2. The main di!erence
A persists for as long as 200 h after magnetic treatment in composition is that the acicular crystals contain less
(Fig. 1). Electron micrographs of carbonate deposits from Mg and no Mn. Neither contains detectable amounts of
Fe ((0.1 wt%).
A comparison of A for 32 pairs of samples, each includ-
2 Manufactured by San Huan, Peking. ing an untreated control is summarized in Table 3. It is
 J.M.D. Coey, S. Cass / Journal of Magnetism and Magnetic Materials 209 (2000) 71}74 73

Table 2 thalpy, but also the lower entropy. Despite its metastabil-
Typical microprobe analyses of calcite and aragonite from min- ity, aragonite formation is favoured at lower evaporation
eral water and well water (at%) rates and higher temperatures [4]. Minute concentra-
tions of cations such as Fe2` [5] and Zn2` [6] can
Well water Mineral water in#uence nucleation. Our data indicate that the magnetic
"eld somehow promotes nucleation of aragonite as the
Calcite Aragonite Calcite Aragonite
water #ows past the magnets; the nuclei are stable for
Na 1.5 2.8 3.1 0.8 hundreds of hours and they grow into the observed
K 0.3 0.2 1.0 0.1 crystals when the water is heated to supersaturation.
Mg 3.7 1.0 3.0 } From the volume of the aragonite crystallites,
Ca 90.7 95.8 90.7 99.1 +3]10~16 m3, the number of nuclei is estimated to be
Mn 3.2 } } } +108/l.
Cl } } 1.7 } The problem is to explain how a magnetic "eld in#uen-
S 0.6 0.2 0.5 } ces nucleation, and why it favours aragonite. Conceiv-
ably, the "eld might:
(i) lower the energy of a nucleus because of a di!erence
in susceptibility with the surroundings. Microprobe anal-
ysis found no iron or manganese in the aragonite crystals,
but even assuming that S"5 ions are present at the
Table 3 2
E!ect of magnetic treatment; numbers of tests in which aragon- 1 at% level in nuclei, energies involved in a 0.1 T "eld are
ite increased (C), decreased (B) or did not change (%) only of the order of 10~2 J/mol.
(ii) in#uence clusters of iron or manganese hydroxide
d AC A% AB SAT SAT that act as heterogenous nucleation centres. For example,
6/ 53
d@FeOOH has a plate-like morphology with a net mo-
Well water 20 14 3 3 0.31 0.50 ment when an odd number of ferromagnetic layers are
Mineral water 12 11 1 0 0.30 0.58 coupled antiferromagnetically [7]. The energy (1) MB of
2
superparamagnetic clusters will be of the order of
Total 32 25 4 3 0.31 0.53
1 J/mol.
(iii) modify the local ionic concentrations via the
Lorentz force q*]B. By analogy with the Hall e!ect,
assuming v"1 m/s and B"0.1 T, the nonelectrostatic
evident that the average SAT tends to be greater in "eld of 0.1 V/m is associated with a surface charge den-
53
treated samples of both waters, although standard devi- sity of 10~11 C/m2. This corresponds to an extra ionic
ations are large because of the variability of the data concentration in the micromolar range, provided the
being averaged (di!erent #ow rates and incubation charge is concentrated in a surface layer 1 nm thick. This
times). The null hypothesis, that magnetic treatment has is still three orders of magnitude less than the Ca concen-
no e!ect on A, can be tested directly on the pairs of data tration, but it is comparable to the concentrations of
on treated and untreated samples where all other condi- OH~ and HCO~ , which limit formation of the CO2~
3 3
tions remained the same. The probability of this is 4% for ion [1]. Statistical #uctuations or turbulence may en-
the well water, and 0.3% for the mineral water. Taking hance it locally.
both data sets together, we deduce that magnetic treat- In conclusion, we have established that a magnetic
ment increases the amount of aragonite in the carbonate "eld e!ect exists. Passing water through a magnetic "eld
deposits, at the 99.9% probability level (3.4 p con"dence subsequently favours formation of aragonite rather then
level). calcite in our experiments, and the in#uence of the treat-
In order to try to understand the mechanism, we recall ment persists for more than two hundred hours. Further
that the reaction experiments on ultra-pure calcium carbonate solutions
are needed to test the hypotheses regarding the mecha-
Ca2`(aq)#2HCO~(aq) nism by which the magnetic "eld produces the e!ect.
3
PCaCO (s)#H O(l)#CO (aq),
3 2 2
Acknowledgements
has an associated Gibbs free energy *G"!24 kJ/mol,
but the free energy di!erence between pure calcite and The work was supported by Forbairt, the Irish science
aragonite at 253C at 1 bar is only !1 kJ/mol and technology agency (SC/96/771). We are grateful to
(+120 K/ion) [1]. Calcite has the lower free energy Dr. Vincent Young for the water analyses and to Dr.
under ambient conditions; aragonite has the lower en- Joseph Taillet for a helpful discussion.
74 J.M.D. Coey, S. Cass / Journal of Magnetism and Magnetic Materials 209 (2000) 71}74

References [5] R.E. Herzog, Q-H. Shi, J.N. Patil, J.L. Katz, Langmuir
5 (1989) 861.
[1] W. Strum, J.P. Morgan, Aquatic Chemistry, Wiley, New [6] P.P. Coetzee, M. Yacoby, S. Howell, Water SA 22 (1995)
York, 1970. 333.
[2] J.S. Baker, S.J. Judd, Water Res. 30 (1996) 247. [7] G.B. Koch, C.A. Oxborrow, S. M+rup, M.B. Madsen,
[3] K. Hihashitani, A. Kage, S. Katumura, K. Imai, S. Hatade, A.J. Quinn, J.M.D. Coey, Phys. Chem. Minerals 22 (1995)
J. Colloid Interface Sci. 156 (1993) 90. 333.
[4] G. Beruto, D. Giordani, J. Crystal Growth 84 (1987) 679.