Вы находитесь на странице: 1из 8


8, AUGUST 2003

Physical Limitations of Antenna

Wen Geyi, Member, IEEE

AbstractA discussion of minimum possible antenna quality plane is [1]

factor ( ) is first presented in the paper. Based on the spherical
wave function theory, the maximum possible ratios of gain to
for both directional antenna and omnidirectional antenna are ex- (2)
amined and new upper limits for both cases have been obtained.
Universally applicable numerical results and some related discus- where is the first associated Legendre polynomial, and
sions are also included in the paper.
is the quality factor of th modes and is a function of ,
Index TermsAntenna quality factor, maximum ratio of gain to with being the wavenumber and the radius of the smallest
quality factor, small antenna. sphere enclosing the antenna. Chus theory is only for an omni-
directional antenna and is based on the equivalent ladder net-
I. INTRODUCTION work representation of the wave impedance of each mode and
the stored energy in some elements has been neglected. Hence

T HE study of physical limitations of antenna can be traced

back to Chus work published in 1948 [1]. Chu has de-
rived the minimum possible antenna quality factor (denoted ),
the above limit just holds approximately.
An analysis of the maximum ratio of gain to antenna for a
directional antenna has been given by Fante [17]. By assuming
maximum gain, and maximum possible ratio of gain to for a that the maximum field strength occurs at ( , ) in
linearly polarized omni-directional antenna using the spherical a spherical coordinate system and the radiated field has only a
wave function expansion outside the smallest sphere enclosing -polarized component, he has obtained the following compli-
the antenna. A lot of related studies have been done since then cated upper limit ([17, eq. (20)])
[2][19]. The antenna is the ratio of the time averaged stored
energy around the antenna to the radiated power and is gener-
ally defined as

(1) (3)

where ;
where and are time averaged stored electric and mag- ; is the associated Legendre poly-
netic energy around the antenna and is the radiated power. nomial; is a constant; and are defined in [3] and
A general study of antenna was given by Fante in 1969 [3]. they are functions of only. The above limit is, however, not
Fantes approach is based on the spherical wave function ex- correct. Actually the maximization of the ratio of gain to
pansion for the electromagnetic fields outside the smallest cir- is an optimization process subject to certain constraint, which
cumscribing sphere of the antenna, along with Collin and Roth- comes from the conditional nature of the definition of antenna
schilds idea that the stored energy can be obtained by the differ- . In Fantes approach this constraint is totally ignored. To
ence between the total field energy and the asymptotic limit of pinpoint the error in Fantes analysis, we cite the expression
the energy density at infinity [4]. The antenna resulting from for the ratio of gain to from Fantes paper as follows ([17,
the spherical wave function expansion method is lower than the eq. (17)]):
real value since the stored energy inside the sphere has been to-
tally ignored in the calculation. Therefore the so obtained is
actually the minimum possible value or the lowest theoretical
limit and any energy stored inside the sphere will increase the
. where and are the coefficients of the spherical ex-
A reasonable quantity characterizing the antenna would be pansions of the fields and
the product of antenna gain and bandwidth, or the ratio of an-
tenna gain to antenna for a high antenna as the antenna
bandwidth is inversion of antenna [5] if is very high. Chu
has shown that the maximum ratio of gain to on the equatorial
Now the coefficients and will be adjusted to maxi-
Manuscript received March 12, 2002; revised June 19, 2002. mize the ratio of gain to . Assuming (thus the upper
The author is with the Research In Motion, Waterloo, ON N2L 3W8, Canada
(e-mail:gwen@rim.net). expression of (4) is assumed to be larger than the lower one and
Digital Object Identifier 10.1109/TAP.2003.814754 should be chosen), Fante has shown that the ratio of gain to
0018-926X/03$17.00 2003 IEEE

will reach maximum if the coefficients and satisfy

((19) in [17])

(5) (7)
In deriving the above equation Fante did not put constraints where the subscript denotes the transverse fields (only the
on and , which is necessary since the antenna transverse field components are explicitly needed in the fol-
is defined conditionally. In other words, and must lowing), and { , } are orthonormal vector basis func-
be confined to a permissible region such that the assumption tions defined by and
will hold during and after the optimization process. , in which is the unit vector in the radial
Otherwise and might fall outside the permissible re- direction, or for and re-
gion to yield , leading to an illogic result. To check if spectively, ,
the optimized coefficients given by (5) are still within the per- , and or 2 for
missible region after optimization, we can substitute (5) into
and respectively, . In
(11) and (12) in [17], which gives the stored electric and mag-
netic energy as follows: (7) indicates the derivative of with respect to
its argument, is the spherical Hankel function of the
second kind. The radiated power, the stored electric energy and
magnetic energy outside the sphere are found to be


By direct numerical calculations, it can be found that

, which contradicts the assumption that . This fact
can be easily seen analytically from the above equation when
is very small since in this case [3]. Therefore,
and have fallen out of the permission region after optimiza-
tion and Fantes limit is incorrect. For example a simple calcula-
tion shows that and for (8)
, and the stored magnetic energy is about 100 times
higher than stored electric energy in this case. Similarly, if the
where and are electric and magnetic field respectively, and
lower expression of (4) is chosen Fantes approach will also lead
is the dielectric constant.
to a contradiction. Furthermore (5) shows that the and
modes are not equally excited under optimized condition, which
means either or modes will occupy a dominant posi-
tion for an arbitrary directional antenna, leading to a situation, Because its clear physical implication, the quality factor has
which cannot be physically explained. been an important parameter in various branches of physics. An
In this paper we will examine the maximum possible ratio of antenna with high will produce a large amount of stored en-
gain to when the antenna geometry and excitation are arbi- ergy around it, which then results in high heat loss and means
trary. As the antenna is of fundamental importance, a discus- narrow bandwidth. So in most cases a low antenna is pre-
sion of minimum possible antenna will be introduced first. In ferred. Assuming that the antenna lies within a smallest circum-
Section III we will derive the maximum possible ratio of gain scribing sphere of radius , the general expression for antenna
to for both directional antenna and omni-directional antenna can be obtained from (8) as [3]
based on the spherical mode theory. It will be shown that Chus
limit in (2) can be pushed to a higher value for an omni-direc-
tional antenna while the maximum possible ratio of gain to
for a directional antenna is totally new.
Since we will use the spherical wave function theory exten-
sively in the paper, some basic relations will be cited here for
reference. In a spherical coordinate system ( , , ), the electro-
magnetic fields outside a sphere enclosing an arbitrary antenna where ; . The
can be represented by spherical wave functions propagating ra- plots of and show that they are strictly monotonically
dially outward [e.g., [20]] decreasing functions of . Some important properties of
and are (1) ; (2) ; and (3)
. Now to minimize by adjusting and we assume
that the first expression of (9) is the largest. To insure that the
first expression of (9) is always larger than the second during
the optimizing process, we need to impose a constraint on the

coefficients and . This can be achieved by assuming can be expressed as

. Under this condition, we obtain

where since .
The right hand side of the above expression can be minimized by
setting or . Therefore the minimum
possible is found to be


which can be achieved by setting and

. If the second expression in (9) is the largest the exactly same Note that all the field components should be included when
result can be obtained by interchanging and . Therefore, dealing with an arbitrary current distribution. For simplicity we
the antenna will attain the lowest if and modes can choose in the following. The radiation intensity is
are equally excited. The existence of a lower bound for antenna found to be
means that the stored energy around antenna can never be
made to be zero or arbitrarily small. If the antenna only radiates
modes or modes, then either or will be zero, and
in both cases it can be shown that the minimum possible is
given by .


It follows from (8), (9), (11) and (12) that
In most applications we need to maximize antenna gain and
bandwidth (i.e., to minimize the for a lossless high an-
tenna) simultaneously. Therefore the optimization of the ratio
of the gain to is more important from the practical point of
view. Obviously the optimization of the ratio of the gain to
will yield a greater minimized than the minimum possible (13)
discussed previously since it demands the gain to be maxi-
mized at the same time. In this section we assume that the an-
tenna geometry and excitation are arbitrary and we will derive
the new upper limits for both omni-directional antenna and di-
rectional antenna in this general situation. Since antenna is
defined conditionally as shown in (1) the optimization of ratio
of gain to is subject to certain constraints.

A. Maximum Possible Ratio of Gain to for a Directive

We assume that the antenna is placed in a spherical coordinate (14)
system ( , , ) and enclosed by the smallest circumscribing
sphere of radius . By definition the directivity is
Only and contribute to the numerator of (13), so the
ration in (13) can be increased by setting for
. Thus (13) can be written as

Now, the spherical coordinate system can be oriented in such a

way that the maximum radiation will be in the direction.
It is easy to find that only contributes to the field in
direction. In this case the transverse fields in the direction of (15)

where the new notations are defined by . Thus we obtain the following upper limit
, , ,
, and (20)

Under the above-optimized condition, the (minimized) and

(maximized) gain are found to be

Here, we use the superscript min to indicate the fact that
We note that the denominator of (15) depends only on the mag-
has been actually minimized with the constraint that the antenna
nitudes of and . If we adjust the phase of and such
gain must be maximized simultaneously. Similarly a superscript
that they are in phase to maximize the numerator the denomi-
max is used to indicate that has been actually maximized
nator will not change. Therefore (15) can be rewritten as
with constraint that the antenna must be minimized at the same
time. Note that the minimized antenna is generally higher
than the minimum possible in (10), since the latter is obtained
without any constraints. The exactly same results as (20) and (21)
can be obtained if the lower expression of (16) is the largest.
B. Maximum Possible Ratio of Gain to for an
Omni-Directional Antenna
Now, we assume that the upper expression of (16) is larger than Next, we assume that the antenna pattern is omni-directional.
the lower one. To guarantee that the upper expression of (16) Since the field is independent of , the vector basis function in
is always larger than the lower expression in the optimizing (7) can be chosen as and
process, we can apply the constraints that , . The transverse fields produced by the an-
, which is similar to what we did in obtaining tenna can then be written as
minimum possible . Under this condition the ratio in (17) can
be maximized by setting and .
Thus, we obtain

Again, all the field components should be included for arbi-
trary antenna geometry and excitation. In the following we let
from Schwartz inequality. In the above,
. The ratio of gain to for the omnidirectional antenna
and ,
may be found to be
with and
, are vectors in Euclidean space
consisting of all vectors of infinite dimension; and
are the inner product and the corresponding norm
defined by


The ratio of gain to attains the maximum if ,



where is an arbitrary constant. The above condition shows that

the and modes must be equally excited to achieve the
maximum possible ratio of gain to , which is in agreement
with the condition for minimizing . From (18) the maximum The first term in the numerator of (22) represents the con-
possible ratio of gain to for a directional antenna will be tribution from the modes and the second term represents

the contribution from modes. Similarly only and Under the optimized condition the corresponding (mini-
contribute to the numerator of (22), so the ration in (22) can be mized) antenna and the (maximized) gain are given by
increased by setting for and
. Introducing the new notations ,
, we obtain

respectively. It should be notified here that all the series in the
above converge absolutely. Therefore there is no need to artifi-
cially truncate them as it has to be in optimizing antenna gain
(24) [1], [16]. In Chus approach the quantity did not occur since
he only considered either or modes. Actually if the an-
tenna is only linearly (vertically or horizontally) polarized then
We note that the denominator of (23) depends only on the either or will be zero in (25), and in this case, we ob-
magnitude of and . Thus the denominator is not changed tain
if the phases of and are adjusted to maximize the ratio.
Evidently, if we choose the phases of and to be the nega-
tive of , the terms in the numerator will be added in phase.
Thus, we have
which is slightly different from Chus approximate limit in (2).
This difference comes from the approximations made in Chus
theory [1], [3]. Since is very close to for most these
two limits should be very close as well. In addition, the new limit
(25) we obtained here is generally higher than Chus limit. Especially
If the first expression of (24) is the largest the above ratio will when is very small we have , thus

i.e., the new limit can be twice as much as Chus limit for a small

Similarly, to guarantee that the first expression of (24) to be IV. FUNDAMENTAL LIMITATIONS OF SMALL ANTENNAS
always the largest during the optimizing process we may assume
With the miniaturization of electronic devices, the reductions
that . Under this condition the above ratio
in antenna size and profile are being continually demanded. A
can be maximized by setting . In this case we have
question that is frequently asked is how small an antenna can be
made while at the same time to maintain a good performance
(26) (i.e., the highest gain and bandwidth at same time). This ques-
tion, which has never been fully studied before, can be answered
based on the above theory.
where , , For a small antenna with , the above-optimized limits
, and can be greatly simplified. In this case we have
. The ratio will reach maximum , and only the first term in the series will
if or dominate. The optimized quantities obtained in previous section
may be approximated as


From (26), the maximum possible ratio of gain to for an (30)

omnidirectional antenna is . Thus, the new upper limit
for an omni-directional antenna is given by

(28) (31)

Fig. 1. Minimum possible Q and minimized Q.

It can be seen that for a small antenna the maximum ratio of minimized will be depicted below. The plots of ,
gain to and minimum possible can be achieved simultane- and are shown in Fig. 1 and they are all strictly monoton-
ously. For both directional and omni-directional antennas, the ically decreasing functions of . In general we have
minimized is the same as the minimum possible obtained , which means that an optimized directional an-
in Section II. It should be notified that (31) are also approxi- tenna is likely to have higher stored energy than an omni-direc-
mately applicable for large (see Section V). tional antenna of the same size. It can be seen that the difference
The last expression of (31) is actually the directivity of an between them is insignificant, especially when is relatively
infinitesimally small dipole. Therefore an omni-directional an- small . In addition we have , which
tenna that can achieve the broadest bandwidth will have the proves that the second of (31) also holds for large .
same directivity as an infinitesimally small dipole, which agrees The plots of and are shown
with Chus theory [1]. It should be pointed out that an infinites- in Fig. 2 and they are all monotonically increasing functions
imally small dipole itself has an extremely narrow bandwidth of . It can be seen that is always larger than
since its real would be much higher than the minimum pos- . The reason for this is that a directional antenna
sible (mathematically it should be infinity). Also note that has the potential of achieving a higher gain when is kept
, which is physically reasonable. about the same level.
The above relationships are the best overall performances an Fig. 3 shows the plots of and . Again they are
small antenna can achieve and they can be used to determine the all monotonically increasing functions of . Note that
smallest possible antenna size once the required antenna band- increases very slowly with , which demonstrates that the last
width is given and vice versa. For example, let us consider an expression of (31) approximately holds for large .
omnidirectional antenna. From the above for- The above curves are universally applicable and very useful.
mulas we may find on the equatorial plane They can be used to determine the best overall performances
and , which is approximately equal to the once the maximum antenna size is given or to determine the
lowest possible value of . Therefore the maximum possible smallest possible required antenna size to get the best overall
fractional bandwidth for a omnidirectional antenna will be performances as we have discussed in Section IV for small an-
. If the operating frequency is 900 MHz, tennas.
this means that the maximum possible absolute antenna band- It is well known that there is no mathematical limit to the
width will be 81 MHz. On the other hand if the operating fre- gain that can be obtained from currents confined to an arbitrary
quency is 900 MHz and the required absolute bandwidth is 100 small volume. But a small sized antenna with extremely high
MHz we have or . From the gain will produce high field intensity in the vicinity of the an-
second expression of (31), we obtain that the antenna size must tenna, which results in high heat loss or high stored energy. By
be greater than . artificially truncating the spherical wave function expansions of
the fields to the order , Harrington has shown that the max-
imum gain obtainable is [16] . Although
Harrington obtained this result by considering a linearly polar-
ized source it can be easily proved that this result generally holds
To see the best antenna performances when is large, the for an arbitrary current source. Hence as increases (equiva-
maximum possible ratio of gain to , the maximized gain and lently the antenna complexity increases) the maximum gain in-

Fig. 2. Maximum ratio of gain to Q.

Fig. 3. Maximized gain and normal gain.

creases. Since the magnitude of the spherical Hankel function gain than while the is kept very small. Actually it can
decreases very slowly for and very rapidly for , be seen from Fig. 3 that if is relatively small both and
the approximate transition point can be considered to will be larger than the normal gain while the antenna is
be the point of gradual cutoff [19]. The normal gain is often in- approximately equal to the minimum possible value as shown in
troduced and defined by letting [1], [16], [18], [19], Fig. 1. Therefore the above statement that the normal gain is the
i.e., . Any antenna having larger gain than maximum gain achievable without incurring high is not true,
the normal gain has been called supergain antennas. It is be- and a more reasonable definition for the normal gain is needed
lieved that the supergain antenna will result in high and there- to maintain its original meaning. It is evident that the value of
fore is not very practical and the normal gain is the maximum should be specified in the definition of the normal gain, which
gain achievable without incurring high . The plot of is is then determined by how much we are willing to tolerate in
shown in (7). various applications.
Obviously the definition of the normal gain or supergain is
kind of ambiguous since the cutoff point is an approxi- REFERENCES
mate transition point and in addition the is not clearly speci- [1] L. J. Chu, Physical limitations of omni-directional antennas, J. Appl.
fied. Indeed it is possible to build an antenna, which has higher Phys., vol. 19, pp. 11631175, 1948.

[2] R. C. Hansen, Fundamental limitations in antennas, Proc. IEEE, vol. [18] R. F. Harrington, On the gain and beamwidth of directional antennas,
69, pp. 170182, Feb. 1981. IRE Trans. Antennas Propagat., vol. 6, pp. 219225, 1958.
[3] R. L. Fante, Quality factor of general idea antennas, IEEE Trans. An- [19] , Time Harmonic Electromagnetic Fields. New York: McGraw-
tennas Propagat., vol. AP-17, pp. 151155, 1969. Hill, 1961.
[4] R. E. Collin and S. Rothschild, Evaluation of antenna Q, IEEE Trans. [20] W. Wasylkiwskyj and W. K. Kahn, Theory of mutual coupling among
Antennas Propagat., vol. AP-12, pp. 2327, 1964. minimum-scattering antennas, IEEE Trans. Antennas Propagat., vol.
[5] W. Geyi, P. Jarmunszewski, and Y. Qi, Foster reactance theorems for AP-18, pp. 204216, July 1970.
antennas and radiation Q, IEEE Trans. Antennas Propagat., vol. AP-48,
pp. 401408, 2000.
[6] R. E. Collin, Stored energy Q and frequency sensitivity of planar
aperture antennas, IEEE Trans. Antennas Propagat., vol. AP-15, pp.
567568, 1967.
[7] , Minimum Q of small antennas, J. Electromagn. Waves Appl.,
vol. 12, pp. 13691393, 1998. Wen Geyi (M00) was born in Hunan, China, in
[8] J. C.-E. Sten, A. Hujanen, and P. K. Koivisto, Quality factor of an December 1962. He received the B. Eng., M. Eng.,
electrically small antenna radiating close to a conducting plane, IEEE and Ph.D. degrees in electrical engineering from
Trans. Antennas Propagat., vol. AP-49, pp. 829837, May 2001. Xidian University (Formerly Northwest Telecom-
[9] A. Shlivinski and E. Heyman, Time domain near field analysis of short munications Institute), Xian, China, in 1982, 1984,
pulse antennasPart 2: reactive energy and the antenna Q, IEEE Trans. and 1987, respectively.
Antennas Propagat., vol. AP-47, pp. 280286, Feb. 1999. He was a Lecturer in the Radio Engineering
[10] D. M. Grimes and C. A. Grimes, Power in modal radiation fields: limits Department of Southeast University from 1988 to
of the complex Poynting theorem and the potential for electrically small 1990; an Associate Professor in the Institute of
antennas, J. Electro. Waves and Appl., vol. 11, pp. 17211747, 1997. Applied Physics, University of Electronic Science
[11] Q
, Radiation of dipole-generated fields, Radio Science, vol. 34, and Technology of China, from 1990 to 1992; a
no. 2, pp. 281296, 1999. Visiting Researcher in the Department of Electrical
[12] D. R. Rhodes, On the stored energy of planar apertures, IEEE Trans. and Computer Engineering, University of California at Berkeley, from 1992
Antennas Propagat., vol. AP-14, pp. 676683, 1966. to 1993; a Full Professor at the Institute of Applied Physics, University of
[13] , Observable stored energies of electromagnetic systems, J. Electronic Science and Technology of China, from 1992 to 2002, where he
Franklin Inst., vol. 302, no. 3, pp. 225237, 1976. was also the Vice Chairman, from 1996 to 1997, and the Chairman, from
[14] , A reactance theorem, Proc. Royal Society of London, Series 1998 to 2000; a Visiting Professor in the Electrical Engineering Department at
A (Mathematical and Physical Science), vol. 353, no. 1672, pp. 110, the University of Waterloo, from February 1998 to May 1998. Currently, he
1977. is a Senior Scientist at Research In Motion, Waterloo, ON, Canada. He has
[15] H. A. Wheeler, Small antennas, IEEE Trans. Antennas Propagat., vol. authored two books and has more than one hundred publications. His current
AP-23, no. 4, pp. 462469, July 1975. research interests include applied mathematics and physics, microwave theory
[16] R. F. Harrington, Effect of antenna size on gain, bandwidth, and effi- and techniques, and antennas and propagation.
ciency, J. Research National Bureau of StandardsD. Radio Propaga- Dr. Geyi is a Member of the editorial boards of IEEE TRANSACTIONS ON
tion, vol. 64D, no. 1, Jan.Feb. 1960. MICROWAVE THEORY AND TECHNIQUES; Journal of Electromagnetic Wave
[17] R. L. Fante, Maximum possible gain for an arbitrary ideal antenna with Propagation (China); Science and Engineering Computation Series; and a
specified quality factor, IEEE Trans. Antennas Propagat., vol. AP-40, Council Member of the China Society of Computational Physics. He is listed
pp. 15861588, Dec. 1992. in Whos Who in the World and Whos Who in America.