THEORY OF THE ACOUSTIC INTERFEROMETER

FOR PLANE WAVES*

by F. E. BORGNIS

Gordon McKay Laboratory of Applied Science, Harvard University, Cambridge, Mass., U.S.A.

Summary
A detailed investigation is presented of the electric input impedance of a piezoelectric
bar and plate, loaded at one face with the acoustic impedance of an interferometer path,
and free at the opposite face. The calculations start from the basic piezoelectric equations
of state and lead rigorously to the strict expression for the total electric input impedance.
The various properties of this impedance are discussed for variable path length of the
interferometer, for variable acoustic wavelength and for a variation of the driving electric
frequency. Under ordinary experimental conditions the input impedance can be repre-
sented by means of circle diagrams. The theoretical results are illustrated by numerical
examples for water at 15 Mc/s and air at 1 Mc/s.
The theory presented provides a rigorous foundation for the well known theory of the
acoustic interferometer given by HUBBARD and extends it in various directions. The treat-
ment includes the basic equations needed for the investigation of highly absorptive media.

Sommaire
On presente une etude detaillee de l'impedance electrique d'entree d'un barreau et
d'une plaque piezoelectrique, charges d'un cote par l'impedance acoustique d'un trajet
interferometrique et libres de l'autre cote. Les calculs sont faits a partir des equations
piezoelectriques fondamentales et conduisent a une expression rigoureuse de l'impedance
electrique totale d'entree. On examine les differentes proprietes de cette impedance en
fonction de la longueur du trajet dans l'interferometre, de la longueur d'onde acoustique
et de la frequence electrique d'excitation. On peut representer l'impedance d'entree par
un diagramme circulaire, dans les conditions experimentales habituelles. Les resultats
theoriques sont illustres par des exemples numeriques relatifs a l'eau (15 MHz) et a l'air
(1MHz).
La theorie presentee constitue une base rigoureuse pour la theorie bien connue de
HUBBARD relative a l'interferometre acoustique, et peut etre etendue dans differentes direc-
tions. L'expose concerne aussi les equations fondamentales necessaires pour etudier les
milieux tres absorbants.

Zusammenfassung
Die elektrische Eingangsimpedanz eines piezoelektrischen Generators (Stab und Platte)
wird ausfuhrlich diskutiert; die eine Seite des Generators ist frei, wahrend die andere
durch die akustische Impedanz einer Interferometerstrecke belastet ist. Ausgehend von den
piezoelektrischen Grundgleichungen wird ein exakter Ausdruck fur die gesamte elektrische
Eingangsimpedanz hergeleitet. Ihr Verhalten wird fur veranderlichen Reflektorabstand,
veranderliche akustische Wellenlange und veranderliche Generatorfrequenz untersucht.
Unter den iiblichen experimentellen Bedingungen laBt sich die Eingangsimpedanz durch
Kreisdiagramme beschreiben. Die Ergebnisse werden an zwei Beispielen fur Wasser bei
15 MHz und Luft bei 1 MHz erlautert.
Die vorliegende Theorie gibt der bekannten Theorie des akustischen Interferometers
nach HUBBARD eine strenge Grundlage und erweitert diese in verschiedenen Richtungen. Sie
enthalt auch die Ausgangsgleichungen zur Untersuchung von Medien hoher Absorption.

PARTI
GENERAL EXPRESSIONS FOR THE ELECTRIC INPUT IMPEDANCE
OF A PIEZOELECTRIC BAR AND PLATE, LOADED AT ONE SURFACE
, surface is parallel to the radiating face. The width
^ of the acoustic beam is assumed so large in compa-

The acoustic interferometer is regarded as consist-
ing of a piezoelectric crystal which is free at one face
and radiates from the opposite face into the inter-
ferometer path containing the fluid or solid medium
whose properties are to be studied (Fig. 1 ) . The
interferometer path ends at a plane reflector, whose
* This paper is essentially based on work done
under a contract between the Office of Naval Research,
Washington, D.C., and the California Institute of Tech-
™*0|£ g ^ ^ p ^ u ^ ^ t he p l p e / w a s ma^e
p 0 8 s i b l e by support of the Division of Engineering and
Applied Science, Harvard University.

152 F.E.BORGNIS: ACOUSTIC INTERFEROMETER
rison with the acoustic wavelength X that the wave
front can be treated as plane. The piezoelectric cry-
stal is driven by an alternating voltage V of angu-
lar frequency oi — 2nf, the current to the crystal
being / . The response of the interferometer to the
driving voltage is completely characterized by the
electric input impedance Z . Indeed, if Z is known,
the response of the device in any driving electric
circuit can be calculated.
M with the length 1. The ratio wjt is assumed to be
f R
— I —* d »•
Fig. 1. Schematic outline of the acoustic interferometer.
C = piezoelectric crystal of length I, M = me-
dium filling the path of length d, R = plane re-
flector, V,l = voltage and current at crystal
input.
The reaction of the acoustic wave-motion in the
m e d i u m upon the crystal can be expressed in
terms of the acoustic input impedance Z m of the
interferometer path as measured at the boundary
between crystal and medium. Our first step there-
fore will be to calculate the electric input impedance
Z of a piezoelectric crystal loaded at its radiating
surface by an acoustic impedance Z m . The opposite
surface of the crystal is assumed to be free (non-
radiating) .
The crystal is assumed to have the shape of either
a thin bar, whose breadth and thickness are small
in comparison with the length, or a plate, whose
linear surface dimensions are large in comparison
with the thickness. The bar is regarded as driven in
compressional lengthwise vibration, the plate in
compressional thickness vibration.
These assumptions are those commonly made in
dealing theoretically with piezoelectric bars or pla-
tes. The waves in the interferometer path are treat-
ed as plane, which implies that the dimensions of
the radiating cross-section are large in comparison
with the wavelength. In the case of a plate this con-
dition is well met. In the case of a bar, however,
only vibrations of high-order harmonics would meet
the t w o conditions that the lateral dimensions are
small in comparison with the length of the bar, and
large in comparison with the wavelength. If the bar
,(1) 1/(2)
A—i—
w tf»
-+/./
y
(2)
Fig. 2. Piezoelectric bar of length / , breadth w and
thickness t, driven by a uniform electric field Ex.
ACUSTICA
Vol. 7 (1957)
is driven in its fundamental mode, the admissibility
of assuming plane waves will depend upon experi-
mental conditions and the nature of the desired re-
sults.
2. Electric input impedance of a piezoelectric bar
in lengthwise vibration with an acoustic load
at one face
Both the breadth w and the thickness t of the bar
(Fig. 2) are supposed to be small in comparison
large enough to ensure a practically uniform electric
field Ex within the bar. The electrodes are assumed
to cover the entire lateral faces of area A = wl. All
surfaces except that at y = l, which is adjacent to
the interferometer path, are free. Consequently all
stresses except Tyy = T2 can be assumed to vanish
everywhere ( [ 1 5 ] , p. 6 1 ) ; T2 vanishes at y = 0.
Only displacements parallel to y, and the compres-
sional strain S2 , need be considered.
The appropriate equations of state for a quartz
X-cut bar with length parallel to a y-axis are 1
S2 = s22,T2 + d12E1, (1.1)
D1 = d12T2 + e^E1. (1.2)
E
For the present purpose we write q = \js22 \
H = d12/s22 effective piezoelectric stress-constant;
V — Ex t is the driving voltage, assumed to be sinusoi-
dal in time. E1 is uniform throughout the bar, but
the electric displacement Dt, while uniform in the
z-direction, varies in the y-direction owing to T2{y).
Equations (1.1) and (1.2) may now be written
as follows, with subscripts omitted, for any point y
along the length of the bar:
T = qES-HV/t, (1.3)
E T
D = HT/q + s V/t. (1.4)
Since these equations are linear and V is sinusoi-
dal, the instantaneous values of T, S, and D are
sinusoidal in time as well as being functions of y.
In what follows, time-dependence will be represent-
ed by the factor &mt, which itself is suppressed.
The equation of motion, in Lagrangian coordi-
nates, for a plane compressional wave with displace-
ment rj in the ^-direction, is
or, since d/dt = j co ,
1
See [6],
— o>2Qcr] = dT/dy.
p. 67 and [7], p. 580; also [2]. For
(1.6)
other crystals than quartz suitable changes in sub-
scripts can be made ([6] and [7]). Numerical values
for quartz are given in Appendix A.
ACUSTICA
Vol. 7 (1957) F. E. BORGNIS: ACOUSTIC INTERFEROMETER 153

From eqs. (1.6) and (1.3), of the interferometer path, which is defined in the
T HV usual way by
CO2 Qc Zm = p(l)/u(l), (1.14)
dy ~E+qB-t
p(l) representing the excess pressure and u(l) the
32r <"2 PP.
, <0< Qc ^rj, (02QC HV
particle-velocity at the boundary between crystal
dy2 (1.7) and medium. Both pressure and particle-velocity
Introducing y2 = to 2 £e/<7£ » w e g e t must be continuous at this boundary. Since a stress
is usually called positive when extensional, whereas
dy t a pressure is called positive when compressional, we
(1.8)
In general yc is complex. The losses in the crystal write p(l) = —T (1). Furthermore, u = drj/dt = jcor).
are included in the complex stiffness qE. They are Thus the boundary condition at y = I can be written
treated in the same way as the losses due to visco- T(l)h(l) = -}o>Zm. (1.15)
sity in fluids.
The expression for Zm will be computed in Part III,
Setting qE = qc0 + jtoF, (1.9)
Section 1. From eqs. (1.6) and (1.12) r\ is found
where qQo is real, we have to be
CO2 Qc
2 _
(1.10) *?=- Facos(ycy + '&),
qc0 + ]coF
For s m a l l losses (w F <^ qcQ), we may write and, introducing <p = ycl, we obtain by eqs. (1.12)
and (1.15)
a sin (99 + $ ) — H V/t j to
X 9c0 V 2qr c0/
a cos (cp + &) /,. qE
or, introducing the crystal phase velocity c c =
V^co/Qc a n ( l , as usual, a space-attenuation constant It is convenient to introduce here the dimensionless
a, we have quantity
zm = a>Zm/ycqE. (1.16)
a) 2 Thus we obtain
7c j a = " j •a (i.ii)
c,. A, HV
Although the crystal losses are treated as small, asin(<£H-#)= + }azmcos(cp+ &). (1.17)
still eq. (1.11) can be regarded as correct for large
By developing sin(9? + $ ) and cos(99 + $ ) in terms
losses also; but in this case the phase velocity c c
of ®™ cp and *™ # and replacing a sin # by means
and the acoustic wavelength Ac in the crystal are
of eq. (1.13), we find by a short calculation
functions of to and F/qc0, as is found by taking the
square root of eq. (1.10). H V 1 — cos cp — j zm sin cp
a cos v = (1.18)
Further losses can arise if imaginary components t sin cp — j zm cos cp
of H and eT are taken into account. Within the
range of frequencies observed so far, no such ima- In order to establish a relation between the cur-
ginary component of H — due for instance to re- rent / and the voltage V, which will lead to the elec-
laxation effects — seems to be indicated and we tric input admittance Y, we integrate eq. (1.4)
therefore regard H as real. A complexe sT would with respect to y :
1 1
imply dielectric losses in the crystal. For quartz, at
least, these losses are extremely small. We shall D Tdy + eT~ (1.19)
treat sT as real; the imaginary component could be / *-?/
0 0
introduced if necessary.
The displacement-current / through the crystal is
The solution of eq. (1.8) for T can be written given by
T = asm(ycy + &)-HV/t. (1.12) 1 1

The constants a and # are determined by the boun- I = w I -^-dy^jcow j Ddy. (1.20)
dary conditions. At the free surface y = 0 the stress 6 0
T vanishes. From (1.12) we have therefore From (1.12) we find
asm& = HV/t. (1.13) 1

At the opposite surface y = I the crystal is in con- Tdy= — [cos # ( 1 —cos cp) 4- (1.21)
/ 7c
tact with the medium in the interferometer path. 0
The reaction of the medium upon the crystal can be + sin "& sin <p] HVl
characterized by the acoustic input impedance Z m
 ACUSTICA
154 F.E.BORGNIS: ACOUSTIC INTERFEROMETER Vol. 7 (1957)

Replacing a s i n # and a c o s # by eqs. (1.13) and
(1.18) and inserting (1.20) and (1.21) into (1.19),
we find finally, with yc = <pll,
pending on the piezoelectric and elastic coefficients
of the particular crystal; they are disregarded in the
present treatment. It will therefore be assumed that
associated with the strain St, only the single stress
7 \ need be considered.

.2 ai AH2V tancp/2-]zm/2
cp qE t 1 — j zm cot cp
The input admittance Y of the crystal is defined by
I=YV and is found immediately from this equa-
tion to be
y-jc^+jc^f'W2-^/2 (1 . 22)
<P q* t l — j z m cot cp
(1)

W he r e Ct-e{l-*fyt. (1.23) Fig. 4. Piezoelectric plate of thickness I, driven by an

The admittance Y can be represented by the electric
circuit shown in Fig. 3. Ct is the parallel capaci-
tance of the crystal, with an effective permittivity
*
i 1_J 1 Owing to the polarization produced by the strain
A B
Fig. 3. Equivalent electric circuit for the loaded piezo-
electric crystal.
The impedance of the branch in parallel with C,
is given by
. <pqEt \-)Zmco\cp
J
2 co AH2 tan cp/2-jzm/2' [
' ]
Eq. (1.24) is the rigorous expression for the electric
input impedance (or admittance) of a piezoelectric
bar, loaded at one end with an acoustic impedance
Zm (expressed by zm). By including imaginary parts
of the quantities cp, qE, sT, and H in eqs. (1.22),
(1.23) and (1.24), all conceivable losses in the
crystal can be taken into account.
electric field E,r.
The equations of state for a quartz Z-cut plate
then become ( [ 7 ] , p. 5 8 0 ; [2])
r^cf^-e^, (1.25)
D1 = enS1 + es1El. (1.26)
Sl(x), the electric field Ex is a function of x over
the thickness / of the plate. The electric displace-
ment Dt, however, is independent of x, as no f r e e
space charge exists within the plate. We write eqs.
(1.25) and (1.26) in the simplified form
T = qES-HE, (1.27)
s
D = HS + e E. (1.28)
These equations for the plate are similar to eqs.
(1.3) and (1.4) for the bar. The numerical values
of the constants, however, are in general not the
same as for the bar (see Appendix A ) .
Proceeding in the same way as in the case of the
bar, we shall establish the differential equation for
T. It is desirable to use the electrical parameter D,
which is independent of x, instead of E. By elimi-
nating E between eqs. (1.27) and (1.28) one finds
Mi+.v)s-7D- (1 29)
-
Introducing

3. Electric input impedance of a piezoelectric

plate in thickness vibration with an acoustic load
*" = ? E ( 1 + ,v) (1 30)
-
at one face we have

T = qOS-~D. (1.31)
The piezoelectric plate (Fig. 4) is assumed to be es
infinitely extended in the y- and z-directions. The By a procedure analogous to that used in deriving
acoustic wave motion is treated as a plane problem eq. (1.8), we obtain
(d/dy = 3 / 3 z = 0 ) , ' t h e only displacement $ being
in the z-direction and the only strain being St. In %i + r<rT=-7c2 SD, (1-32)
a f i n i t e plate other strain-components, together oar £3
with all six stress-components, may be present, de- where, in the case of the plate, yt.2 = o;2 Qc/qD.
ACUSTICA
Vol. 7 (1957) F. E.BORGNIS: ACOUSTIC INTERFEROMETER 155

Applying the same boundary conditions 2 as in eq. Now, by eq. (1.30)

(1.15) at x = l and the condition T = 0 at x = 0, we es qE H2 ESqD
H+
obtain the solution for T corresponding to eq. (1.32) : H "+„,r
HV es J H
H and we find, introducing the current / = j co D A ,
r = asm{ycx + &)-" D. (1.33)
v== 11 _ 2H2U tan rp/2 ~jzm/2
s 2
The two equations determining the constants of inte- j o) A e j <p co A (ES) qD 1 - j zm cot rp
gration a and # are found from the boundary con-
ditions; they are
where Z, the input impedance of the crystal is
a sin & = D (1.34)
ES
* - . \ ( i - . 2 f D t a r / 2 _ i W 2 ) d-42)
and
]OiCi\ (pesqD 1 - j z m cot cp )
a sin (rp + # ) = ,. # + j « «m cos (rp + i>), (1.35) with Ci = Aes/l. (1.43)
According to eq. (1.42) the impedance Z can be
where
represented by the equivalent circuit of Fig. 5,
zm = w Z
; and <p = ycl. (1.36) where
Z,- V 2 f „ t a i n ^ 2 - i - ^ 2 . (1.44)
From eq. (1.35) it follows by means of eq. (1.34) 0) Gj (p E* qu 1 — j zm cot rp
that Eq. (1.42) gives rigorously the electric input im-
„ H D 1 - cos <p - j zm sin 99 n „_. pedance of a plate in thickness vibration; as in the
a cos v = . (1.-1/J case of the bar, all conceivable losses are included,
£* sin 9? — j 2 m cos rp
if <p , qn, H, and fs are regarded as complex quan-
On eliminating 5 between eqs. (1.27) and (1.28) tities.
we find A B

T--(ff+?V)* + «y ) (1.38)
Fig. 5. Equivalent electric series circuit for the loaded
and integrating eq. (1.38) with respect to x, we piezoelectric plate.
have
Although Fig. 5 corresponds to the simplest ana-
, lytical expression for the impedance of the crystal
fTta--{H+' f)r+^'. (1.39, plate, as given by eq. (1.42), it is of advantage in
0 many cases to transform the circuit of Fig. 5 into
On the other side we find from eq. (1.33) the more usual form shown in Fig. 3. It can easily
be shown that the two circuits are equivalent in that
they represent at all frequencies the same input im-
Tdx= [a cos ^ (I-cos rp) +
7c pedance or admittance with respect to the points A
(1.40) and B, if the following relations exist:
+ a s i n # sin 9?J — 1+jwCjZ,
1 (1.45)
o)2Cl2Zl 1

Inserting a s i n # and a c o s # from eqs. (1.34) and

(1.37) we obtain after a short calculation Inserting Zj from eq. (1.44) and Ci = C 1 from eq.
(1.43) into eq. (1.45) we obtain after a short cal-
culation
J yces l-)zmcot<p e*
o Z> = / 9 ° ' • (1-46)
Introducing this result into eq. (1.39), we get 2 c) A H£
esqE Dl •S „E , H2 tan rp/2 ., /<n /., t 2 H2 \
H+ H+rf-\- '-ESqo j ( W 2 ) o t
H H W2 - r ^ - ^ v ) .
j tan rp/2 + zm/2
2HD Um<p/2-jzj2
As already stated, the expression for Zx is more
yces 1 - j z m cot rp complicated than that for Z\. For the crystals com-
monly used, however, the quantity | H2/es q° | is
2
The metallic coating on the plate is assumed to be small in comparison with unity; this fact permits
so thin that its acoustic impedance is negligible. a simplification of eq. (1.46).
 ACUSTICA
156 F. E.BORGNIS: ACOUSTIC INTERFEROMETER Vol. 7 (1957)

PART II
ELECTRIC INPUT IMPEDANCE O F THE FREE PIEZOELECTRIC
BAR AND PLATE
1. The free bar As seen from eq. (2.2) the real part of Zx is a func-
tion of oj; this means that R^ varies with Ao) =
The formulas for a f r e e crystal, that is in ab- oj — OJs. But as a I is assumed to be small, terms of
sence of any external load, follow immediately from order a I • Aw can be ignored near resonance, and
the previous general formulas. If the surface at R\, can be regarded as a constant as is commonly
y = l is free, the stress T (I) vanishes. Therefore in the case.
eq. (1.15), the condition that T(I) = 0 is equivalent
The quantities Lh and C|, for the bar in Fig. 6 are
to Z m = 0 and zm = 0 according to eq. (1.16) for a
found from eqs. (2.1) and (2.3) by using the rela-
bar and (1.36) for a plate. Thus we obtain the im-
tions 4
pedance Zj for the free bar, corresponding to the
upper branch of the equivalent circuit in Fig. 3, by 1 dZ t . _ 1
Ih = and Ch= (2.6)
setting zm = 0 in eq. (1.24) : 2 dco

7 . gEt <P w After short calculations one finds easily

Zu= COt (2.1)
-*a>'A& 2 2-
b
Usually this impedance is represented in the neigh- 8 A H2 c2 (2.7)
bourhood of the series resonance cos by the circuit 8 AH2 1
shown in Fig. 6. Series resonance occurs when the and cb n2qEt (2n + l ) 2 '
imaginary part of Z1 vanishes, or, in terms of C
and L , when OJ2 C L = 1 . in agreement with the well-known values of these
quantities ( [ 5 ] , p. 2 9 7 ) .
With
, (o I . , A n t i r e s o n a n c e is here defined as occurring
when the imaginary part of the total admittance Y
becomes zero, that is, when Im (j 0) Cx + l/Zj) = 0
and regarding the losses in the crystal as small so
(see Fig. 3 ) . Disregarding terms of order higher
that a I <^ 1 , we have
than a I, the angular frequency coa for antireso-
w w CO I . al al nance follows from eqs. (2.1), (2.2), and ( 1 . 2 3 ) :
£ cot ^ = J « ;ot J
2 2 2 c, 2cc oo^ I OJ.XI H2
c
°t ~' (2.8)
or, retaining only first-order terms in a 1/2 and de- 2 c, 2 c, e' qc
veloping the cot-term, Assuming that H2/eT qE *^ 1 , and moreover that
(p <p wl o)l aI I oil coa 1/2 c c is very close to (2 n + 1) Jt/2 , we can write
„ cot -~ = -—- cot ---— j •«- (2.2)
2 2 2cc 2cc 2 ILcot
""2cc
°^cl = ( 2 n + l ) g + <$„ {H2/eTqE<\).
to 1/2 c c
sin 2 (coZ/2 c c ) + Then after a short calculation it is found from eq.
The imaginary part of Z1 therefore vanishes when (2.8) that
cot(oj//2c c ) = 0
or cos 1/2 c c = (2 n + 1) JI/2, (2.3) ^ , 0 ^ i f T E<^ (ffV*V<D. (2.9)

where n = 0, 1, 2, 3, • • •. Thus, as is well known, ( 2 / 1 + 1) ite1 qL

s e r i e s r e s o n a n c e for the bar occurs at From eqs. (2.3), (2.9), and (2.10) it is seen that
= (2n+l) ncjl. (2.4) coa — c o s « 2 c c 7 = T P .„ *., „ • (2.10)
a s c
I eTqE ( 2 n + l ) 2 J t 2
The real part of Zx , representing the resistance R\> This is the difference in angular frequency between
of the bar at resonance, follows from eqs. (2.2), antiresonance and series resonance for a free bar.
(2.3), and (2.1) 3 : For the usual crystals it is a very small quantity
qEtl (see Appendix A . l ) .
( * b ) „ _ =aZ (2.5)
4>AH2cc
4
Since Zx is represented in the neighbourhood of re-
3
The consideration of small imaginary parts of qE sonance by Z1=jo) Lb + 1/j o)Ch + Rb, dZjdoj = j (Lb +
or H would only lead to corrections in s e c o n d order + l/co2 Cb), and since cos2 Lb Cb = 1, | dZjdoj |c=a,.,
terms and can therefore be omitted. = 2 Lb.
 ACUSTICA
Vol. 7 (1957) F. E.BORGNIS: ACOUSTIC INTERFEROMETER 157

In the foregoing treatment, as well as in what fol- Thus series resonance for the plate occurs a t 6
lows, the effect of the electric circuit (matching net-
TCCC 4CC • H2
work, generator, etc.) is disregarded. So far as the w s= (2/i+l) (2.15)
I ( 2 / i + l ) nl esqD
use of a crystal in the interferometer is concerned,
the theory shows that only crystal-resonance is of (H2/esqD<l).
interest. Therefore no heed will be given to the cha-
The real part of Zx, representing the loss-resistance
racteristics of the external electric circuit.
Rv (Fig. 6) for the plate, is found in the same way
as for the bar. From eqs. (2.11), (2.12), (2.14),
and using the development eq. (2.2) for small los-
2. The free plate
ses, we obtain by a short calculation, assuming
a I < 1 and dp < 1
The impedance Zx corresponding to Fig. 3 for
this case is obtained from eq. (1.46) if we set
zm = 0 . We find (*p) al (2.16)
iAH*c~
H2
Zlc = ] q2n COt (2.11)
mAH \2 2 *s»D

The term H2/es qD may be attributed to the effect of

polarization, which makes the electric field Ex vary Fig. 6. Representation of Zx in Fig. 3 near resonance by
along the x-direction (Fig. 4 ) . Such an effect does C, L and R in series.
not occur in the bar, where the electric field is per-
pendicular to the direction of mechanical vibration. The quantities Z,p and Cp for the plate, corres-
As already mentioned, H2/es qD is a small quantity ponding to the equivalent network of Fig. 6, are
for the crystals commonly in use. When this term is obtained by using eq. (2.6). Neglecting terms in
neglected, the expression for Zx in eq. (2.11) has dp, we have
the same form as the expression for the free bar in q°P
p
eq. (2.1). 8AH2c2
(2.17)
S e r i e s r e s o n a n c e for the free plate occurs,
_^AW- 1
according to eq. (2.11) when 5 and p
~ %2qDl "(2/1+1) 2 '
<P H2
Re cot s D 0 (2.12) in agreement with the well known values of these
eq
quantities ( [ 5 ] , p. 3 2 3 ) .
For small losses (a I <^ 1) we make use of the same A n t i r e s o n a n c e of the free plate occurs when
development of ((p/2) cot(9?/2) that led to eq. (2.2). I m ( l / Z 1 + j o ) C 1 ) = 0 . From eqs. (2.11), (1.43),
Assuming that the imaginary parts of qD/H2 and of and (1.45) we obtain
es are small quantities, we find the condition for
.9? cp
series resonance to be R e I -«r- c o t -n = 0
0)J 0)sl H2
- — cot - — = Ke (2.13)
2, cc Z cc ; 5 qD and for small losses, r e p l a c i n g ^ c o t ^ byeq. (2.2),
A simple approximate solution of this equation fol- we find
lows under the assumption that Re(H2/es qD) < 1 ,
coJ/2cc= ( 2 n + l ) jt/2 (2.18)
by setting coa 1/2 c c = (2 n + 1) (rt/2) — Sp , where Sp
is a small quantity. Retaining only first order terms From eqs. (2.15) and (2.18) it is seen that for a
in dp and ignoring an imaginary part of H2/ES qD, free plate the difference in angular frequency be-
we find from eq. ( 2 . 1 3 ) : tween series resonance and antiresonance is
(2 cjl) dp, where dp is given by eq. (2.14) when
H2/esqD<l. Thus7
( 2 n + l ) Jt e* qD
H2 4 to
5
Eq. (2.12) is identical with the result obtained by esqD (2n+l)2jt2 (H2/esqD<l).
a different approach by R. BECHMANN in his theory of (2.19)
the vibrating plate ([1], eq. 41). BECHMANN'S corrective
6
term in qD (c —c in his notation, eq. (32 a ) ) , however, It is easily shown that eq. (2.15) agrees with
is too great by a factor 2, due to the omission of a fac- BECHMANN'S expression for series resonance, as given in
tor 1/2 in his expression for the piezoelectric inter- [1], also in [5], p. 314.
7
action-potential ip (in BECHMANN'S notation); see also Eq. (2.19) agrees with BECHMANN'S results, [1],
R. BECHMANN [2 a ] . eqs. (61) and (61 a ) .
 ACUSTICA
158 F. E.BORGNIS: ACOUSTIC INTERFEROMETER Vol. 7 (1957)

PART III
MECHANICAL AND ELECTRICAL IMPEDANCE
OF THE INTERFEROMETER PATH
1. Mechanical impedance of the interferometer negative wave at the reflector is given, according to
path eq. (3.7), by

The mechanical input impedance of the interfero- ("--) =r=\r\j&. (3.8)

meter path at the radiating boundary of the crystal
is by definition, eq. (1.14), In general T is complex; the modulus | T | indicates
the diminution of the reflected wave-amplitude with
_T{~d) p(-d)
(An) x = (3.1) respect to the incident amplitude, and is associated
u(~d) u(-~d)' with the energy absorption in the reflector. The
if we choose the coordinate system of the interfero- phase angle & indicates the "phase-jump" which
meter path as indicated in Fig. 7. T, p , and u denote the reflected wave undergoes with respect to the in-
stress, excess pressure, and particle velocity of the cident wave. For a perfectly rigid reflector (a) a r = o
medium in the path. R is a plane reflector parallel is zero by definition, and therefore J 1 = — 1 (or
to the crystal boundary. The differential equation | r | = 1 and 0 = ji) 8 .
The stress T = qm d£/dx is, according to eq. (3.4),
-*1R
T(x) = -jym qm £0 (e~j ^ - r e i V m X ). (3.9)
x—d K-IQ
Thus the mechanical input impedance Z m , as defin-
Fig. 7. Simplified sketch of interferometer with crystal ed in eq. (3.1), is found by using eqs. (3.4) and
C at x — — d and reflector R at x = 0 . (3.9) at x= -d:
for a plane compressional wave motion, in Langran- (3.10)
gian coordinates (displacement £ ) , is j CO I CO l + f P
2jy
m rf
3 2 32£ (3.2)
_i l
Introducing the dimensionless quantity
Qm
~dfi dx2
-2jy m d
•=r. (3.11)
where Qm is the density and qm the stiffness of the
medium. To include the losses we assume qm to be we have, at x
complex: 1-r
ZJm — (3.12)
qm = 9m0 + j co Fm . (3.3) CO l"+r
The solution of eq. (3.2) is The dimensionless quantity zm in eq. (1.16) for the
bar and eq. (1.36) for the plate becomes
£ 0 r ) = M e ~ J V + re3ym*), (3.4)
i2
1-r
2
n , ri_ 7 - -'J2J» (3.13)
where v - -J«D CO< (3.5) 7c 9c 7c ?c 1+7 •
where qc = qE is the stiffness of the l?ar, and qc = qD
The phase velocity of the plane waves according is that of the plate. We introduce
to the solution eq. (3.4) is, by definition, c m =
to/Re (ym). If the losses in the medium are assumed m = IUQ -f- j m^ = (3.14)
to be small, c m can be expressed by ycVc

C m 2 = ?mo/ftn ( w ^m < ?mo) • (3,6) 8

The quantity T in eq. (3.4) is the reflection coeffi-
cient of the reflector R . The particle velocity
u = dtj/dt = j co £ at the reflector (x = 0) is given by
eq. ( 3 . 4 ) :
(u)x-_Q = jo)i0(l+r). (3.7)
If we construe the two terms in eq. (3.4) as belong-
ing to two wave components progressing in the posi-
tive and negative z-directions, the ratio of the par-
ticle velocity u+ of the positive wave to a . of the
The suggestion is sometimes found in the litera-
ture that some kind of "reflection-losses" may occur
also at the boundary between crystal and medium and
that such losses might affect the value of r. There is
no room, however, for such considerations within the
scope of the theory given here. The boundary condi-
tions at the interface between crystal and medium re-
quire continuity of pressure and displacement in the
two adjacent media. Nothing e,lse has to be introduced.
The situation would be different if dissipation by heat
transfer or surface layers due to impurities were taken
into account; but such effects are beyond the scope of
the theory treated here.
ACUSTICA
Vol. 7 (1957) F. E.BORGNIS: ACOUSTIC INTERFEROMETER 159

1-r The first terms in eqs. (3.18) and (3.19) corres-

and have zm = m (3.15) pond exactly to the expressions for Z± found for the
l+r
free bar in eq. (2.1) and for the free plate in eq.
From eq. (3.5) and the expression preceding eq. (2.11). The second terms may be construed as re-
(1.8), the quantity m can be expressed by presenting that part of the total electric impedance
Zx which is linked with the interferometer path.
m = y.njlto- = lAm_^m = 1 /^m ^mO ( 1 + j CO F m / g m 0 )
Writing Zt as the sum of the input impedance of
7cgo X Qcqc V Qcqcoii + i^'fc/qco) the free crystal Zi c and the impedance of the inter-
Qm C*n "I / l + j (o
Cur m /y m 0
Fm/q* ferometer path j?ii, that is Z± ±= Z^ + Z^, and intro-
m — m0 + j m1 =
\ 1+jaiFcfqco ducing <p = yc I we find
(3.16) 1
Zn = zQ (3.20)
2 . <p
If the losses in both crystal and medium are small, 10Ot
we can write
i- 2
where
Qm Cn
m = m0 + j m t =
Qc ^c
1+i
2-(^-r)«
2 WmO qcO/.
(*o) bar
- 2wA~lP ~v- qEQ* u__ (3.21)

where
and
Qm Cm 2AH2
Qc c c (co F m < qm0) (*o)
0/ plate (3.22)
(3.17)
(om0(Fm Fc (co J1,. < c/c0)
The equivalent circuit of Fig. 3 for the loaded
2 \<7m0 f/cO c o , 4 # 2 by vqthat shown
crystal can be 2replaced Q
*2AH2 in Fig. 8;
The quantity mx can also be expressed by introduc- Z l c is the impedance of the f r e e crystal, which may
ing the space-attenuation parameter a , as defined be represented in the neighbourhood of resonance by
by eqs. (1.11) and ( 3 . 5 ) : the jRLC-branch in Fig. 6, with the values of R> L
and C as given in Part II.
_ " * o Vcm a m — cc ac) • (3.17a)

2. Electrical impedance associated with the

interferometer path
Fig. 8. Equivalent electric circuit for the loaded piezo-
The impedance Z1, as given in the general form electric crystal. Zi c impedance of the free crys-
of eqs. (1.24) and (1.46) for the bar and the plate, tal, Zii impedance of the interferometer path.
can be thought of as containing two parts: one be-
longing to the driving crystal, the other to the load, Our main interest lies in the impedance Z^\, as
that is to the interferometer path, as represented in represented by eq. (3.20). This impedance is given
our formulas by the quantity zm. It is possible, by the same expression for the bar as for the plate,
indeed, to split the expression for Z± into the sum but z0 might have different values in the two cases.
of two such parts. It can be verified easily that Zt, Inserting zm from eq. (3.15), we obtain
according to eqs. (1.24) and (1.46), can be written 1
7 (3.23)
as follows:
a) B a r :
1 1+ r j cot
<P
m 1—r
qEt 1 <p 9 +
Z
^^AWjfCOt2 (3.18)
3. Circle diagram for the electric impedance Z^
E of the interferometer path
qt <pj2
+ co AH2 "2" ~ <p The usual purpose of the acoustic interferometer
^ - j c o t ^
is the determination of the velocity c m of sound and
b) Plate: the absorption coefficient a m of the medium filling
qDl \(<p H2 the interferometer path. This is accomplished by
*i« 2 cot ~sr — s D
w + (3.19) changing the distance d or the acoustic wavelength
co A H j \ 2 * 2 e q
Xm in the medium and observing, for example, the
, aDl W2 corresponding changes of the current / to the crys-
+ wAH22 . w
— •— j cot -j- tal (Fig. 1) at constant voltage V. All changes of
zm 2 parameters of the matching electric circuit connected

160 F.E.BORGNIS: ACOUSTIC INTERFEROMETER
with the crystal are associated with the changes of
the electric input impedance Zi = Zic + Zu of the
loaded crystal (Fig. 8 ) . The behavior of Z l c be-
longing to the f r e e crystal has been treated in
Part II. In all cases except when the frequency is
varied, Zj c is constant, because it is independent of
the variables of the interferometer path. Our main
interest lies in the variation of Z ^ , associated with
the interferometer path.
Eq. (3.23) gives Z\\ as a function of r and
cp = yQil. First we note that, according to eq. (3.5),
ym can be written for any medium as
(3.24)
Cm
Then, according to eqs. (3.11) and (3.8),
r = p e~2i(odlc™ e~
2a d
n (3.25)
j(6-4nd/X ) ~2amd
•=\r /? e~J*, (3.26)
where
-2a„
R= \r\-e
and 0 = 1 ^ - © = 2J°A -6. (3.27)
The variation of r is effected either by a) changing
the pathlength d at constant acoustic wavelength Xm
(variable path interferometer), or by b) changing
Xm at constant d (fixed path interferometer). The
reflection coefficient | T | eJ@ is treated here as a
constant, independent of the frequency.
Case a ) :
The variation of d usually is of the order of magni-
tude of a few half-wavelengths 2 m /2 . Each change in
d by Am/2 changes the term exp (j • 4 n d/lm) in eq.
(3.26) by 2 jt and causes a periodic change in r by
a full period. The term exp( — 2 am d) in eq. (3.26)
decreases continuously with increasing d. If a m is
not unusually large, this term can be regarded as
"slowly" varying in comparison with the "rapidly"
varying preceding factor in eq. (3.26).
Case b ) :
The term exp( — 2amd) in eq. (3.26) is cons-
tant; the periodic changes in r are effected only by
an alteration of the number of half-wavelengths
within the interferometer path, either by a change
in the sound velocity c m (due to changes in pres-
sure, temperature or composition of the medium)
or by a change of the driving frequency / .
The first step in the graphical representation is
to consider the locus of r. In the complex r-plane
r = R e - ^* describes' exactly a c i r c l e with centre
at the origin, when Xm is changed steadily (Fig. 9 a ) .
A change of d, however, causes r to describe a
s p i r a l , due to the factor exp ( — 2 a m d) in eq.
ACUSTICA
Vol. 7 (1957)
Fig. 9. Locus in the r-plane of the quantity r = Re~^
in eq. (3.26) with variation of (a) the acoustic
wavelength Am and (b) the path length d. In
(b) the ratio between two successive radii is
given by Rn+1 = Rne~a™1™ •
(3.26). I n w h a t f o l l o w s w e a s s u m e a m
t o b e a r e l a t i v e l y s m a l l q u a n t i t y ; we
ir1 ay then replace the continuous spiral by a se-
qu mce of circles whose diameters decrease by a
faci ~>T exp ( — a m X) each time when r has undergone
a change of one full period (Fig. 9 b ) . Only for
media with unusually large absorption will the spi-
ral contract so quickly that this approximation has
to be replaced by a more exact treatment. This can
be done by using the general expressions of eqs.
(3.23) and (3.26).
We next find the locus of Z^ in eq. (3.23) when
r describes a circle around the origin, as indicated
in Fig. 9. Z\\ in eq. (3.23) can be written in the
following form:
Zyx m 1 m/2
(3.28)
(l+p)2 l_p i+"p
r + ••
l+p
where <P (3.28 a)
P=J cot
The term cpj2 will be regarded as a constant, as is
the case when the driving frequency is kept cons-
tant. Even if the interferometer is used at varying
frequency, d/Xm is usually a large number and
cot 99/2 is a very slowly changing function near crys-
tal resonance (qp^rn) in comparison with the rapid
change of r so that, as a first approximation, cot (p/2
can be regarded as constant in eq. (3.28) (see also
Part V ) . The kind of function connecting the va-
riables Zn and r in eq. (3.28) is called a h o m o -
g r a p h i c or l i n e a r substitution. It is known
that a circle in the r-plane is always transformed
into a circle (or a straight line in special cases) in
the Z^-plane (see, for example, [ 8 ] , p. 605, Sec.
258) 9 .
To obtain the cirCle in the Z^-plane correspond-
ing to the r-circles as indicated in Fig. 9, the follow-
ing steps are to be performed in conformity with
eq. (3.28) [ 8 ] :
9
Circle diagrams in connection with the determina-
tion of sound absorption have been introduced also by
C. W.KOSTEN [14].
ACUSTICA
Vol. 7 (1957) F. E.BORGNIS: ACOUSTIC INTERFEROMETER
1. Transformation of the origin 0,- in the r-plane
to a point P = - ( l - p ) / ( l + p ) .
2. Rotation of the plane through 180° around a
line through P whose angle with the real axis
is given by one-half of the angular coordinate
of m / ( l + p ) 2 .
3. Inversion with respect to P of the r-circle in
its displaced position with a constant of inver-
sion equal to \mj(1 +p)2\.
4. Displacement of the original origin 0 r to the
point 0 = ( m / 2 ) / ( l + p ) , which is the origin
for the Zij/zo-circle.
As these steps include only displacements of ori-
gins and an inversion, it is obvious that every
r-circle is transformed into a circle (or straight line)
in the Zu-plane. The position of the centre C as well
as the radius Q of the corresponding Zu-circle can
be calculated by the exact expression eq. (3.28).
This calculation will be given in the following sec-
tion under the simplifying assumption that cot <p/2
is constant and real; this assumption is justified
under the usual experimental conditions. Fig. 10
shows schematically the positions of the r-circle and
the Zu-circle. Since the real part of Z^ is always to
be positive, the Zu-circle must be situated at the
right side of the imaginary axis of the Zu-plane. It
is easy to see that the i n n e r part of the r-circle is
transformed into the i n n e r part of the Zu-circle.
Fig. 10. Corresponding circular loci in the r-plane
(left) for r = R e~^ and in the Zu-plane
(right) for Zu/z0 = x-\-] y . The circles are de-
scribed by r and Zu/z0 as indicated by the sense
of the arrows with increasing &, that is with
increasing frequency.
The greatest radius R occurring in the r-plane
corresponds to the value d = 0, as seen from eq.
(3.27). R equals | T | in this case; /?max = l f ° r a
perfect reflector. Owing to the absorption in the
medium, R decreases steadily when d increases. If
d tends toward infinity, R tends toward zero; d = oo
corresponds to a crystal radiating into an infinitely
extended medium, where no reflected wave exists
(r = 0 ) . The corresponding value of Z\\ follows from
eq. ( 3 . 2 3 ) :
z
_ o
(Zii), (3.29)
9 product a m Xm is relatively large.
j cot
m In the case of the f i x e d - p a t h interferometer,
This value of Zj; represents the electric impedance
equivalent to the radiation-load of a crystal radiat-
161
ing from one face into an infinitely extended me-
dium. The sequence of r-circles with decreasing
radii R is represented in the Zu-plane by a sequence
of circles with decreasing radii Q. But whereas the
centre O r in the r-plane remains fixed, the centre C
in the Zu-plane moves towards the point C 0 given
by eq. (3.29), since the coordinates (xc, y^) of C
are functions of R. Fig. 11 gives a qualitative pic-
ture of the correspondence between circles in the
two planes. Actually, as we saw above, a change in
d causes r to describe a spiral around 0 r , whose
radius decreases slowly if the absorption in the
medium is not extremely large. The corresponding
locus of Zji is also strictly a kind of spiral whose
"centre" is slowly moving as indicated by the se-
quence C 3 — C 2 — Q — C 0 in Fig. 11.
Fig. 11. Location of corresponding circles in the r-plane
(left) and the Zii/z0-plane (right). C0 denotes
the centre with zero radius.
In practical interferometry, u s i n g a v a r i -
a b l e p a t h - l e n g t h d, only a few consecutive
cycles in the r-plane are described, whose radii
change slowly under ordinary conditions. Thus Z\\
describes the same number of cycles in the Zu-plane
with slowly decreasing radii and a small displace-
ment of C , when <xm is not extremely large. A full
cycle in the r-plane is performed when the change
in (4 K d/Xm) equals 2 jt (eq. ( 3 . 2 6 ) ) , or when
A{2djXm)=l. (3.30)
At constant F the initial radius R = \r]e~2a^d
-2a (d+Ad)
changes into R' = | -T | e m in each cycle, or,
using (3.30) at constant An
R'/R = ^ = e ~Vi (3.31)
If o m A m < l , R'^R or Zn' Zu according to eq.
(3.23). In other words, when we start at a certain
distance d corresponding to a certain value of Z\\
and change d into d + Xm/2 , we obtain very appro-
ximately the same value of Z\\, if a m Xm can be as-
sumed to be small. Or, a change of d by Xm/2 causes
Z\\ to describe very approximately a complete cycle
in the Zu-plane. This is not the case, however, if the
where Xm is varied at c o n s t a n t d, the term
e am in eq. (3.27) remains unaltered. Here we
 ACUSTICA
162 F.E.BORGNIS: ACOUSTIC INTERFEROMETER Vol. 7 (1957)

have to deal with only one circle of constant radius _ m0 l+R2

XC
in the /"-plane, corresponding to one circle with cons- ~~2 l'-JR*'
tant radius in the Zjj-plane. A change in Am corres-
m± l+R2 .\m\2 <p ,„„.,
ponding to eq. (3.30) causes Zy\ to describe e x -
a c 11 y one full cycle, provided the term cot (p/2 can *°-Tl-#+ 4 COt
2' (3 34)
'
be regarded as constant in eq. (3.23). This means \m\R
that an increase by one of the number of half-wave-
lengths filling the path-length d leads to exactly the
( c o t y / 2 <€ 1 and real; m^ < m 0 ) .
same value of Z\\.
When cot (p/2 = 6 = 0 , these formulas are valid for
any value of mjm0, as is seen by eq. (3.33).
4. Radius and coordinates of the centre of the
Zji-circle

In the preceding section we dealt qualitatively

with the conformal representation of the r-circle
and the corresponding Zjj-circle. Now we establish
the analytical expressions for the coordinates XQ , y^
of the centre C and for the radius Q of the Z^-circle,
which corresponds to an r-circle with radius R and
centre 0, in the r-plane. In eq. (3.23) we put
ZU/ZQ = w=x + jy ; r = Re~J* ; cot(p/2 = b (assum-
ing b to be a real constant) ; m = m0 + j mt, and
obtain
Fig. 12. Location of circles in the plane of w=x + jy
x + jy- (3.32) = Zn/z 0 corresponding to concentric circles in
the r-plane with different values of radius R.
(mo + j m t H l - f t e - J * )
2(1+R e-J*j - (j m0 - mx) b (1 - R e~J*) Eq. (3.34) gives immediately
Setting the real and imaginary parts of the left side
of eq. (3.32) separately equal to zero and eliminat-
ing cos # and sin # from these two equations, for
J 2 \m I2 ! m I2 fi 2*\
example, by squaring and adding them, we obtain and Q*= •-•-' X Q - (3.36)
the equation for the locus of W = Z1[/ZQ which cor- 7MQ 4
responds to an r-circle with radius R and the origin Thus the centres of the circles for different values
as centre. The result is of R are located on a straight line, whose angle tp

{x-[2m0{l+R2)]lD2}2+{y-[2mx{l + R2)+\m\2b{l-R2)MD2}2

= 4 / n 0 2 ( l + / ? 2 ) 2 + [ 2 7/^(1 + R2) + | m \2 b (1 - R2)I ] 8 _ | m | 2 ( 1 - R2)

where
D2 = 4,(l-R2) +4m16(l+/?2) +m2b2(l-R2).
In most practical cases the quantities cot (p/2 = b
and mJrriQ can be regarded as small (see also
Part I V ) . Ignoring all terms of order b2, mx b , and
higher, we find the following expressions for the
coordinates XQ , yc of the centre C and for the ra-
dius Q of the quantity w = Zyjz0 :
with the ar-axis is given by tan y> = mJmQ. The ra-
dius £ increases with increasing Xc towards Q = oo
at XQ = oo or R = 1 . The smallest possible value for
XQ is given by xc = m0/2 with Q = 0 or R = 0 .
Fig. 12 shows the positions of different circles
for different values of R ; it confirms the more qua-
litative picture in Fig. 11.
ACUSTICA
Vol. 7 (1957) F. E.BORGNIS: ACOUSTIC INTERFEROMETER
PART IV
GENERAL PROPERTIES OF THE ELECTRIC INPUT IMPEDANCE
OF THE ACOUSTIC INTERFEROMETER
UNDER VARIOUS EXPERIMENTAL C O N D I T I O N S
1. Quantitative considerations
Following the more qualitative considerations of
Part III we now give attention to the values of the
more important parameters in the main equation
(3.23), as encountered under practical conditions.
a) T h e q u a n t i t y 7n = /n 0 + j " * i :
The imaginary part mx is very small in compari-
son with m0 under ordinary conditions. Neglecting
losses in the crystal in eq. (3.17 a ) , that is assuming
<xc c c ^ a1T1 cm , and introducing c m = cu0 Xm/2 Jt, we
obtain from eq. ( 3 . 1 7 ) :
m0 2 Jt
The order of magnitude of a m // 0 2 (in c m - 1 s2) is
known to be 1 0 ~ 1 3 in gases and 1 0 ~ 1 6 in liquids 1 0 .
The velocity of sound may be assumed to have
the order of magnitude of c m = 3 • 10 4 cm/s in gases
and c m = 10 5 cm/s in liquids. The quantity a m Xm
may be assumed therefore to be in the order of mag-
nitude of
9 It is therefore advisable to choose the driving fre-
amCm — 3 - 1 0 - / 0 in gases,
"m ^m :
/o m 1 0 - 1 1 ' /o in liquids.
Even up to the megacycle range the quantity a m Xm
can be regarded as small in comparison with unity
(especially in liquids). Still one must be careful in
special cases, where the assumptions made above
may not be justified; for some substances a m and/or
c m may be of larger order of magnitude; then / 0
must be smaller to justify the assumption that
am Xm < 1 .
Therefore, according to eq. ( 4 . 1 ) , m1 can be
neglected ordinarily in comparison with m0 and m
may be regarded as a real quantity | m \^nnQ.
b) T h e q u a n t i t y r = Re~*& :
The foregoing considerations, which resulted in
the estimates for amXm according to eq. (4.2), jus-
tify also the assumption mentioned already in
Part III concerning R, namely that R, according to
eq. (3.27), is a slowly varying function, when d is
steadily changed by an amount of a few half-wave-
lengths. Thus the spiral in the Z^-plane mentioned
in Part III, Section 3 , due to a variation of the path-
length, contracts very slowly with the change of d
and may reasonably be replaced by a sequence of
circles, as outlined.
10
See [3], for example pp. 464 and 538.
163
c) T h e q u a n t i t y cot<p/2 :
According to eq. (1.11) <p = ycl = 2n(l/Xc) — j aQJ;
the crystal is usually driven at or near its fundamen-
tal mode, that is l^Xc/2 and cp/2^. (jt/2) — j ac Xc/2.
The attenuation in crystals is known to be extremely
small, so that we may regard <p/2 as a real quantity,
neglecting the term j a c Xc/2.
In the neighbourhood of <p\2 = jt/2 the cotangent
function can be replaced by (jt/2) AX/X0 , where
AX is defined by X~XQ +AX = 2l + AX in a piezo-
electric plate. Usually X equals 2 I very approxi-
mately and the cotangent term in the denominator
of eq. (3.23) is a small quantity which can be ne-
glected in comparison with the other term. The or-
der of magnitude of mmm^ is 1 0 - 5 for gases and
1 0 - 1 for liquids. Thus in gases the weight of the
cotangent term is considerably smaller when com-
pared with the weight of the first term, which con-
tains the factor 1/m, than it is in liquids. The co-
tangent term may become important when r in eq.
(3.23) comes very close to — 1, that is when
R = | r | e °m< is close to unity.
(4.2) quency so that cot (p/2 equals zero, that is at reso-
nance for the free bar and antiresonance for the
free plate (see Part I I ) . If the interferometer is used
at v a r i a b l e frequency, it will depend upon the
experimental conditions (| T \ and <xm • d), whether
the cotangent term in the denominator of eq. (3.23)
has to be taken into consideration or not.
2. Connection between Z i ; and the formulas of
Hubbard for the acoustic interferometer
J. C. HUBBARD [12] was the first to develop a
theory for the acoustic interferometer. Upon his
theory are based all subsequent investigations with
this device. The difference between his theory and
that given here is that HUBBARD'S method, although
leading by an ingenious procedure quite quickly to
his well-known results, nevertheless makes it diffi-
cult to perceive the significance of the various as-
sumptions involved. On the other hand, the theory
offered here starts from the basic piezoelectric
equations of state and leads in a rigorous and
straightforward way to more general results.
HUBBARD showed that the reaction of the inter-
ferometer load upon the electric input impedance of
the driving crystal can be expressed by an impe-
dance, which for convenience we call
 ACUSTICA
164 F. E.BORGNIS: ACOUSTIC INTERFEROMETER Vol. 7 (1957)

ZJI = const (P - j (?) ,

for the bar,
in series with the impedance of the free crystal. YAH2
const (4.7)
This is, as it should be, quite in agreement with Qm Cm '
2 for the plate.
the foregoing considerations, which led to the equi- YXH
valent circuit of Fig. 8. The impedance Z^ in We may repeat that HUBBARD'S formulas do not
our notation must consequently be identical with include the cases where a) the frequency differs
HUBBARD'S ZH , if the assumptions implicitly involv- from cc/2l, b) the reflector induces a phase jump
ed in his calculations are introduced in our eq. of the reflected wave different from JI , and c) the
(3.23). These assumptions turn out to be the fol- absorption coefficient a m is unusually large. Eq.
lowing: (3.23) is rigorous, covering all possible cases. Still,
1. cot 99/2 = 0 : The crystal is driven exactly at under ordinary experimental conditions, one would
its resonant frequency in the case of a bar find no appreciable numerical difference between
(cf. eq. (2.4)) or at antiresonant frequency in the two formulas.
the case of a plate (cf. eq. ( 2 . 1 8 ) ) . The losses
in the crystal are negligible.
3. Maxima and minima of the impedance Z\\
2. m = m0 ; m1 = 0 : The losses in the medium
are not unusually large, so that m1 can be ne- of the interferometer path and their relations
glected in comparison with m0 (cf. Part IV, to am, \r\, and 0
Section 1 a ) . Assuming ordinary experimental conditions we
3. 0 = JI : The phase-jump of the reflected wave shall describe some characteristic features of the
in the interferometer path equals 180° ex- periodic changes of the impedance Z j ; , as given in
actly (eqs. (3.8) and ( 3 . 2 7 ) ) . eq. (3.23), due to changes of the path length d or
Introducing these assumptions in the general ex- the acoustic wavelength Xm .
pression (3.23), we obtain By ordinary experimental conditions, the follow-
ing is meant:
11 {q> 6)
2 1+Re-V - 1. / 0 = const.
After separating the right side into real and imagi- 2. 99/2 = Jt/2 ; cot 99/2 = 0 . The losses in the crys-
nary parts, we find tal are negligible, so that 99 is real. The cor-
1-R 2 responding frequency / 0 is given by / 0 = c c /2 /;
m0z0
Zu 2 + (4.4) thus 99/2 = nljX = Jt/2 and the cotangent term
2 \l+R + 2Rcos&
in eq. (3.23) vanishes. This means that the
• 2 fl sin #
+ J 2 crystal is driven at its resonant frequency in
L+R + 2RcosJ
the case of a bar, or at its antiresonant fre-
Comparing this expression with HUBBARD'S impe- quency in the case of a plate (see Part II,
dance eqs. (2.4) and ( 2 . 1 8 ) ) .
Z H = const (P-)Q) (4.5) 3. The coefficient of absorption a m in the me-
dium is not unusually large, so that the quan-
and introducing tity amXm is small in comparison with unity
$=(2cod/c„)-jt and R = | T | e~2°md (see eq. ( 4 . 2 ) ) . Then the locus of Z j ; , due to
according to eq. (3.27) with 0 = :t , we get a change in d of amount Am/2 , may be regard-
a d ed as practically a circle. When d remains
1 - iIrri 212, * - 4 -
1
fixed and Xm is changed, the locus of Zj; is an
-iamd exact circle, regardless of the value of a m Xm.
i + |r 2 l e m cos
(4.6) If the assumptions mentioned under 1. 2. and 3.
are valid, Z\\ describes a circle as d or Xm is varied.
0 , r I -2,d . 2 (ad
2 \1 e m sin The radius Q and the coordinates XQ , yc of the cen-
tre C of the circle described by the dimensionless
-4a„ o l r l -2a d 2wd quantity W = Z1\/ZQ has been given in eq. (3.34).
i + |r 2 \ 1 e mm cos
For experimental reasons it may be desirable to
in conformity with HUBBARD'S expressions Px = o and know also the locus of the a d m i t t a n c e Y\\ =
Qx = o (in his notation), as given, for example, by 1/Zn. The corresponding radius Q and coordinates
F. E. Fox ( [ 9 ] , formulas (6) and ( 7 ) ) . The cons- XQ , yc of the centre C of the circle described by the
tant in eq. (4.5) is found by using eqs. (3.17), dimensionless quantity w — z0 Y-^ are easily found
(3.21) and (3.22), and introducing y = <p/l = jc/2 I by inverting the w-circle as illustrated in Fig. 13.
= O)VQc/qc = nccV Qc/qc jl: The inversion is effected by w = \jw or in coordi-
ACUSTICA
Vol. 7 (1957) F. E.BORGNIS: ACOUSTIC INTERFEROMETER 165

nates, according to the notation in Fig. 13, by of m, but depends upon the absorption in the me-
dium. From this ratio the quantity R can be deter-
x = V
(4.8) mined, which involves the absorption of the medium
x£ + yi ~ x2 + y2
(oimd) as well as the losses ( | - f | ) at the reflector.
where x + j y = ifr^ and x+ j y — w. A circle Measurements at two different values of d allow one
(x — XQ)2 + (y — yc)2 = Q2 m the w-plane corres- to separate the losses in the medium from those at
ponds to the circle (x — xc)2 + iy — yc)2==Q2 m the the reflector, as is well known [ 1 0 ] . Eqs. (4.11) are
w-plane; the values of xc, Vc an& l} °f the inverted obtained under the assumption that the change in
circle are found by means of eq. (3.34) with R due to the factor e m is small; therefore they
cot <p/2 = 0 : should be applied to a d j a c e n t maxima and mi-
2m0 1+R2 2mx 1+R2 nima at each of the two path-lengths.
xc yc | m \ 1 - R2 '
2 The coefficient of absorption a m and the coeffi-
\m\2 1-R2'
cient of reflection | T | can be calculated as follows,
4 R
Q = (cot 9?/2 = 0) (4.9) if the ratio | Z1{ | m i n /| Z 1; | m a x or | YVl | m i n /| Yri | m a x is
m 1-R* known for two different path lengths d± and d2 :
Introducing
1-R 7 V I
" l i |min •* l i |min (4.12)
1+R 7 I V I
1
" l i |max li max
and 2 am d — In I 7 1 (4.13)

we have, according to eq. (3.27),

Fig. 13. Loci of the quantities w = Zu/z0 (left) and
w = z0 Yii (right) corresponding to the impe- -2amd _ x

R = \r = e (4.14)
dance and admittance of the interferometer
path. The sense of the arrows corresponds to and from eq. (4.12)
an increase in frequency.
l-e~* , X l+o
1 + e- -- tanh -=- , or X = In
1-0
(4.15)
Fig. 13 shows that the maximal and minimal va-
lues of I w I and \w\, occurring at the points A and Denoting by ot and o2 the values of o measured at
B or B and A , are given at any m = m0 + j m1 , by the two distances dt and d2, we have the relations
w !ax = ]/a£ + VG + Q, I M> Imax \ j / ' ^ + V% + Q> A j = A otm d j •Inlrl-lntd + oJ/U Oi)L
w lin = }/%c + y% — Q, \w|min = fal + yl — Q• X2 = 2 am d2 l n | r | = ln[(l + a2)/(l o-2) L
From eqs. (3.34) and (4.9) we obtain whence we find (4.16)
(l_ai)(l + a2)
z0 I m I 1+R In (4.17)
7 = _2__. i ^ , am = 2(d -d )
2 1 (l + ai )(l_a2)
I " l i |max

V l i. IIm a x
J
2 1+R and r l = e a m ^, + d , ) ] / ( l - a 1 ) ( l -- 0o 2; ) . (4.18)
'0 m\ 1-R (4.10) y ( l + a j a + a 2)
z0\m\ 1—R r\=e2amdi,* 1-0 1.2
II 7" l i I|min
. also (4.19)
~~2~ 1+ R 1 + 0i,2 '
2 1-_R_ When d can be made so large that 2 a m d ^> In | .T |,
V I
-1 l i mm z0\m\ 1+R the following simplified formula may be used, which
The following relations result: is obtained from eq. (4.16), neglecting In | T7 j :
Z 1 l+o
Zu 7 \ — SL In (2amrf>ln|r|). (4.20)
"li min — Yd l-o-
On the other hand if d can be made so small that
Yu -* l i min — (4.11]
z02 I m e "m ^al, \r\ can be measured directly. Eq.
7 I . 1-/? y« i-jg (4.19) then becomes
" l i |min
" l i max 1+/? yu 1+ R •„, l-o
{2amd<l). (4.20 a)
l +o
The p r o d u c t s of the maximal and minimal
values of | Z^ | or | Y-^ | are independent of R, that Returning to assumption 3. at the beginning of
is of I r I and a m d. On the other hand their r a t i o this section, mx is very small in comparison with m0
is independent of the properties of the crystal and under ordinary conditions. This means that yc in

166 F.E.BORGNIS: ACOUSTIC INTERFEROMETER
eq. (3.34) or (4.9) is very small in comparison
with XQ . Thus the centres C and C of the circles in
the w- and w-planes lie practically on the real axis;
the maximal and minimal values of Z\\ or F u are
practically pure resistances and conductances. Under
the condition that m1 is ignored in comparison with
m0, C and C are located exactly on the r e a l axis
and the difference in & between the points A and B
or B and A in Fig. 13 is exactly to Ad' = n ; from
eq. (3.27) it is seen that this corresponds to a va-
riation Ad = n cm/2 OJ = A m /4. Indeed, according to
eq. (4.4). Zn is real when sin # = 0 or & = ± n n ;
in this case \Z^\ reaches its maximum ( z i = l , 3 ,
5 , . . . ) or its minimum (n = 0, 2, 4 , . . . ) . Therefore
the difference in # in proceeding from a maximum
to the next minimum or from a minimum to the
next maximum in | Zn | (or | Yn | of course) is
equal to %. Or, if m1 < m0, a change in path-length
of Am/4 leads from the real maximum of | Zn | or
| F n | to the adjacent real minimum or vice versa.
This is not so, however, if the absorption in the
medium is unusually large; in this case the corres-
ponding change in path-length differs from Am/4
and the maximal or minimal values of the moduli of
Z\\ or Fii occur at c o m p 1 e x values of Z\\ or Yn.
Since for mt ^ m0 the maxima or minima of | Z^ |
or | Yn | occur at # = + n n , we find from eq. (3.27)
that the corresponding values of d = dm are given by
r a
#=±njt= ^ -&
d jacent minimum.
or »-(n+f-)^L (mi<m*)' (4-21)
n = l , 3 , 5 , . . . correspond to a maximum and
n = 0, 2, 4 , . . . to a minimum of | Zn | , The phase
jump 0 of the reflected wave can be determined by
means of eq. (4.21), when the distance dm and the
number of quarter-wavelengths filling the interfero-
meter path are known. The d i f f e r e n c e in dm,
corresponding to a maximum and an adjacent mini-
mum in | Zn | or | Fii | , is always Am/4 , whatever 0
may be, if 0 can be regarded as a constant.
Eq. (4.21) can be applied also to the f i x e d -
p a t h interferometer, where d is kept constant and
the acoustic wavelength lm is changed.
With either type of interferometer, for a perfect
reflector (@ = Jt), eq. (4.21) states that a m a x i -
m u m of | Zji | occurs always when an e v e n num-
ber of quarter-wavelengths, that is a whole number
of half-wavelengths, fills the interferometer path; a
m i n i m u m of | Z\\ | occurs at an o d d number of
quarter-wavelengths in the path. If the phase jump
0 differs from Ji, the number of wavelengths filling
the path at a maximal or minimal value of | Zn I is
no longer a whole multiple of A m /4.
ACUSTICA
Vol. 7 (1957)
The d i f f e r e n c e , however, in the number of
quarter-wavelengths filling the interferometer path,
and corresponding to a maximum and an adjacent
minimum of | Zn | or | Yn \, is always u n i t y ,
whatever the value of the constant 0 (or | T | of
course, which quantity does not enter eq. (4.22) at
all) may be.
4. Determination of the velocity of sound cm
from the maxima and minima
of the impedance | Zn \ n
As is well known, the velocity of sound c m in the
medium can be found from the distance between
maxima and minima of | Zri | (or | Yu | ) . Under the
assumptions 2. and 3 . of the preceding section,
ZIJ/Z 0 describes a full circle in the H/-plane (Fig. 1 3 ) ,
when the change in & amounts to 2 J I (see eqs.
(3.26) and ( 4 . 3 ) ) . Thus the distance between two
adjacent maxima or minima of | Z\\ \ (or \Yx\\)
corresponds to a change
J
U-j-4d-2" (4 22)
-
for constant 0, as seen from eq. (3.27).
If the variation of Zn (or Yu) is caused by a
change of path-length Ad, we obtain from eq. (4.22)
cm= Ad = 2 f Ad (m1<m0), (4.23)
where Ad is the change in path-length necessary to
proceed from one maximum of | Zn | to the adjacent
maximum or from one minimum of | Z\\ \ to the ad-
If the variation of Zn (or Yn) is produced at
constant frequency / 0 by a small change AXm in the
acoustic wavelength ^ m , owing, for example, to a
change of pressure, temperature or composition of
the medium, the corresponding change Acm is found
from eq. (4.22) to be
__Acn^nc m *0m
cm co0d 2d
This is the relative change in c m corresponding to a
change in Am by an amount AA.m = Acm/f0 great
enough to make | Zn \ change from one maximum
to the adjacent maximum, or from one minimum
to the adjacent minimum.
If the variation of Zn (or Yn) is produced by a
change of the driving frequency / = co/2 JT , the ne-
cessary change Af to proceed from one maximum of
| Z I J | (or | Yn \) to the next or from one minimum
to the next is, according to eq. (4.22),
Af = AOJ/2 jt = c m /2 d,
whence the velocity c m is
cm = 2dAf. (4.25)
11
See also [4].
ACUSTICA
Vol. 7 (1957) F.E.BORGNIS: ACOUSTIC INTERFEROMETER 167

A change of 2 jt in & always corresponds to a full radii for R^l to small radii for R < 1; at /? = 0
"period" of Z\\, that is to a closed circle in the the radii g and g are zero (eqs„ (3.34) and (4.9))
w>-plane; it is obvious that the foregoing formulas and the "centres" are at XQ = m0/2 and xc ~ 2/mQ.
(4.22) to (4.25) hold not only for adjacent maxima According to eqs. (4.27) and (4.28) the moduli
or minima of | Z^ | but also for the more general | WD I — m 0 /2 , and | ws | = 2/m 0 , are independent of
case in which one proceeds from an a r b i t r a r y R. The circles with 0 and 0 as centres and radii
value of Z\\ (or Y\\) to exactly the same adjacent m 0 /2 and 2/m0 therefore intersect the w- and u>-circ-
corresponding value. les for different values of R at the points D and D
Still, as pointed out in the preceding section, whose coordinates are given in eq. (4.29).
under ordinary conditions (m1 < /n 0 )
the change in # necessary to proceed from a m a x i -
m u m of | Z\\ | to the adjacent m i n i m u m equals
n, that is just half the amount found necessary to
proceed from a m a x i m u m t o the adjacent m a x i -
m u m in the general case. The foregoing formulas
can therefore be modified for the case in which a
m a x i m u m of | Zu | and an a d j a c e n t m i n i -
m u m are chosen for the determination of cm. The
formulas read in this case as follows:
c m = 4 / ' Ad', (4.23 a)

_ H » (mt <^ m0) (4.24 a)

c
m 4,d
Cm =4d- Af. (4.25 a)
Fig. 14. Location of the points D (impedance) and D
(admittance) for # = 2 n x + n/2 , and of D'
5. On the form of the periodic curves and D' for # = 2 n jt — Jt/2_ in the to- or
for | Z n | or | Yu\ w-plane. The points A or A correspond to
# = (2n + l) JT, the points B or B to # = 2 n r t .
Assuming again m1<€ m0, that is circles in the
w- and u)-planes (Fig. 14) with centres on the real
axis, we now see what happens if # changes, for Fig. 14 can be used to represent the w-plane as
well as the w-plane with the notation given in
example, from ^ B = 0 at B (or B) to # D = ± JI/2 .
Fig. 1 3 ; only the scale-values have to be chosen pro-
This change in # corresponds to a change in d of
perly. Incidentally, if mQ = 2 the scale values are the
± A / 8 , as is seen from eq. (3.27). The correspond-
same for both planes, as seen from eqs. (4.27) and
ing values of w ~ ZH/ZQ and w — z$ Y^ are found
(4.28).
from eq. ( 4 . 3 ) :
From Fig. 14 qualitative curves for (2/m 0 z0) | Z^ |
2 1±jS
«>D _ » ! l ± j « . and —
Wt) or (m0 z0/2) \ Y^ | plotted against # can be found,
(4.26)
2l+]R m0 1+jR
as shown in Fig. 15. If R is close to unity, the w- or
whence ^-circle is large, and D or D (or, correspondingly,
l-/g2 . 2R D ) are much nearer to the minimum B or A than
(4.27)
l+R* + 31+R*
H>D = WB
4 to the maximum A or B of (2//n 0 z0) | Z^ | or
(m 0 z 0 /2) | Yn |. At a n y R the quantity (2//n 0 z 0 ) | Zu \
1-R* . 2R 4 (and also (/ra 0 2 0 /2)| Yn |) has the s a m e value at
(4.28)
m 0 \ l + / ? 2±J %r =
W>D =
- J 1+R* mn # = ± J T / 2 . Thus at R^l the curves drop down
By means of eqs. (3.34) and (4.9) the coordinates rather rapidly from their maximal values to the
of wj) = Xn + jyj) and Wf> = x^ + )yt) can be written value 1 at ^ = Jt/2 and have comparatively flat mi-
as nima.
2R 2R _ When R ^ 1 the circles in the w- or ty-plane are
X\) = XQ Q XT, = XC-
TTR* i + #2' (4.29) small and the points D or D have nearly the same
distance from B as from A (or from A as from B ) .
l~fl2 . 1-R2_
± l + R2 ys ± Thus, as R decreases, maxima and minima assume
T^Q
VD=
more and more closely the same shape. At /? = 0,
Fig. 14 shows the location of w^ and w^, denot- corresponding to a reflector located at infinity or to
ed by D and D , for different values of R (cf. the a perfect absorber, the periodicity vanishes altoge-
notation in Fig. 1 3 ) . The circles vary from large ther, and a constant value of I Zu I or | Yu I is reach-

168 F.E.BORGNIS: ACOUSTIC INTERFEROMETER
ed, namely the value for a crystal radiating into an
infinite medium (eq. ( 3 . 2 9 ) ) .
It may be noted that the curves for (mQz0/2) | Y^ j
are obtained by shifting the curves for (2/m0z0)\Zli\
to the right or to the left by an amount of & = Jt in
Fig. 15, since
[ ( 2 / m 0 z 0 ) | Z l i | ] ^ = [(/n0z0/2)|yli|]^
according to eq. (4.3).
R-3IU
ff=1/3
-'r~1---V--J—ff-0
r\ hi
KV 1+/?
•R- 3 M
ff-1/3
/?=0
© Fig. 16. Total input impedance Zi ••:Zi c + Zii for a
Fig. 15. Representation of (a) | Zii | and
m0 z0
m0z0.
Yn I versus ! ? + 2 n 3 i for different va-
lues of R.
The curves in Fig. 15 permit the following con-
clusions. At the same path-length d and the same
reflection coefficient | F | the curves for | Z\\ | or
| Yij | show sharp maxima and flat minima as long
as the absorption a m is small. With large absorp-
tion this difference between maxima and minima
tends to disappear. Or, for the same medium and
the same reflector, the curves for | Z^ | or | Y^ \ in
the neighbourhood of two different distances d1 and
d2 such that amd1<€l and a m </2 > 1 , show a sharp
maximum and a flat minimum at dt, whereas at d2
this difference disappears. Different curves for | Zy\ \
or | Y\\ | always intersect at whole multiples of
# = ± jt/2 ; the multiplicity can be found by the
number of quarter-wavelengths along the path-
length d.
ACUSTICA
Vol. 7 (1957)
We may add the following remark at this point:
the t o t a l input-impedance Zx of the interferometer
is given by the sum Z^ + Z^, as shown in Section 2
of Part III. In all cases where the crystal is driven
at constant frequency, the constant term Z l c has to
be added to Zx\ . When (p/2 = jt/2 , as we have been
assuming, Z l c is the real "loss-resistance" due to
losses in the crystal and mounting. Z\\ contains the
factor m0z0 (eq. ( 4 . 3 ) ) , and is therefore propor-
tional to mjf2, as seen from eqs. (3.21) and (3.22).
Since m 0 is about 4000 times smaller in gases (at
atmospheric pressure) than in liquids, a t t h e
s a m e f r e q u e n c y the crystal is considerably
more loaded by a liquid than by a gas.
Under ordinary circumstances, that is at frequen-
cies not exceeding the order of magnitude of
1 Mc/s, a gaseous medium contributes essentially
in the neighbourhood of | Zn | to the total impedance
Z x , whereas in the neighbourhood of | ZX\ |min the
value of Z i c is predominant. Fig. 16 illustrates this
case. The distance OB, corresponding to | Zx\ |min » is
only a small fraction of O j O , corresponding to Z J c .
gaseous medium.
The minimum of Zx is chiefly determined by Zlc,
(b)
and is therefore rather flat. The sharp maximum of
Zj is due to the sharp maximum of | Zn |, as shown
in Fig. 15 a, occurring at # = Jt, 3 J I , 5 j t , . . . or
dm = nlm/2 according to eq. (4.21).
In liquids, owing to the much greater value of
m0, the crystal is considerably loaded, up to very
high frequencies, even when the impedance | Zxi |
has its minimal value, because the ratio
is so much larger than in gases. The position of Ot
in Fig. 16 is then rather near to 0 . The behavior of
| Zx | is therefore approximately the same as that of
I Zn I alone (Fig. 15) 12 .
12
See also [9], (Fig. 2 c). The characteristic difference
in the shapes of the curves for current versus d in
gases and in liquids, as experimentally observed and
described in that paper (see also [3], Fig. 186, p. 245)
is a consequence of the characteristic features of | Z\\ \,
as described above and also mentioned by Fox. The
particular curves for the current as given by Fox are
essentially due to the particular circuit connected with
his interferometer and to the frequencies applied.
ACUSTICA
Vol. 7 (1957) F.E.BORGNIS: ACOUSTIC INTERFEROMETER
6. Locus of minima and maxima of | Z i ; | and | Yn |
If the path-length d is steadily increased, the mi-
nima of | Zn | or | Yii | rise and the maxima fall con-
tinuously, both finally tending towards the same
value as d approaches infinity. Assuming the same
conditions as mentioned in Section 3, and making
use of eq. (4.10), we see that the locations of the
minima and maxima of both | Z\\ | and | Yn | are
associated with the function
l-R _ l-exp(-2amc? + lnlr|)
/(</) =
l+R ~ T+exp~(-2omrf + l n | r | j
1
/ (d) = tanhj a m d — _ In (4.30)
Fig. 17. Locus containing (a) the maxima and (b) the
2
t I 7 I J IOTI Z« I V I
minima of .—r— L\\ and — „ — I Y u ver-
| m | z0 A
SUS Ctm d .
PART V
THE VARIABLE-FREQUENCY INTERFEROMETER
For those cases in which movable parts should be
avoided, the fixed path interferometer offers con-
siderable experimental advantages [ 1 3 ] . The change
of | Zjj | at constant path-length can be effected by
a variation of the acoustic wavelength Xm in the
medium, as seen from eq. (3.26). This variation
might take place at constant frequency / 0 , as, for
example, when the pressure, the temperature, or the
composition of the medium is varied. From this
variation the corresponding change Acm of the velo-
city c m can be found, as shown by eq. (4.24). Still,
the absolute initial value of c m must be known in
order to plot the changes Acm against the variable
parameter that causes Am to change.
A determination of c m itself can be accomplished
by varying the driving frequency at constant path-
length. If one could observe the corresponding
change in | Z^ | , c m could easily be found by means
of eqs. (4.25) or (4.25 a ) . Also the absorption-
coefficient am could be calculated by formula (4.17)
169
Indeed, f{d) contains all the minima of
ZYi | • 2/(z 0 | m |) or of | Y^ | • z0\ m |/2 , whereas the
maxima of these quantities are on the reciprocal
curve l/f(d) , as seen from eq. (4.10).
Fig. 17 shows the curves for f(d) and l/f(d)
according to eq. ( 4 . 3 0 ) . The origin 0 is at a di-
stance of (1/2) In | r | from the origin 0 ' of the
curve for / ' = tanh <xm d. If | T | is close to unity
and a m d small, we can set | T | = 1 — £ , where £ is a
small quantity, and obtain from eq. (4.30) as a first
approximation
f(d)^amd+ -£ (4.31)
(amd<l, e<l).
It will be noted that the minima are located approxi-
mately on a straight line, which intersects the ordi-
nate-axis at f{d) = e / 2 = (1/2) ( l - | r | ) . The maxi-
ma lie on a hyperbola l/f(d). This corresponds to
remarks, found in the literature, that the "peak en-
velope" is rather a hyperbola than an exponential.
The envelope is never a simple exponential, as seen
from eq. (4.30).
It may be noted that our statements concerning
maxima and minima are related to | Z^ \ or | Yn
and n o t to the t o t a l impedance | Z± | = | Z l c + Z^
which is necessarily the observed impedance. Whe-
ther or not | Zi c I m a Y be neglected depends on the
actual ratio | Z l c | to | Z\\ | . In many cases, especially
at small values of m and when R comes close to
unity, proper consideration must be given to | Zj c | .
or (4.20). The difficulty here is that the quantity
determined experimentally is | Zx \ = \ Z l c + Z^ | ,
where now Zi c also varies due to the change in fre-
quency.
At constant d the locus of | Z^ | is practically a
circle, whose centre, under ordinary conditions
(/nx <^ m0), lies practically on the real axis. The
reason why this locus is not s t r i c t l y a circle is
seen from eq. ( 3 . 2 3 ) : by changing the frequency
the term cot <p/2, which is zero for the reference
frequency / 0 = c m /2 / , deviates from zero. Accord-
ing to eq. (3.34) the centre C of the circle for Z^
moves away from the z-axis, as the frequency, and
consequently <p, changes. Still, under most practical
conditions the "coupling" term cot (p/2 in the deno-
minator of eq. (3.23) may be kept so close to zero
that it can be neglected in comparison with the term
2 (1 + r) /m (1 - r) (see also Part IV, Section 1 c ) .
Quantitatively we may say that the cotangent term
may be ignored, if
 ACUSTICA
170 F. E.BORGNIS: ACOUSTIC INTERFEROMETER Vol. 7 (1957)

* \AL\ 4 A (i-R)
2 /0 | in

f- 31 m
(1 rie" 2 a m d )'

A more serious complication, however, is involv-

(5.1)

ed by the fact that the crystal gets off resonance

when / is varied. In the neighbourhood of the refe-
rence-frequency / 0 we can make use of the equiva-
lent circuit of Fig. 8 and write Fig. 18. Locus of Zi=Zic + Zn at variable frequency.
As / is changed, the origin 0 moves along the
ZU = RZ + ) L L - - ~ } = - - R C + ]2LACD. (5.2) dotted line normal to the x-axis.

This situation is illustrated by Fig. 18. As / changes,
the origin Ox for the impedance Z\ = Z l c + Z1} moves
perpendicularly to the x-axis. Let 0 / be the location
of O t at Aco = Acox. It is obvious that the maximum
(or the minimum) of Zx now differs by O / E from
the maximum 0 A A (or the minimum OjB) belong-
ing to Ao) =! 0 . (E is the intersection of O / C with
the circle of radius CO\ around the centre C). Fur-
thermore the difference # B ' — $ A ' between the values
of # belonging to B' and A' deviates from n if A'
and B' move away from A and B. Thus both the de-
termination of a m (associated with | Zxi | m a x and
| ^li |min) and of c m (associated with A$ — # B ' — &y)
are affected.
Adequate analytical expressions might be calcu-
lated from eqs. (3.18) or (3.19), giving the neces-
sary corrections for the determination of a m and
c m . This requires, however, more detailed calcula-
tions, which will not be dealt with here. Corrections
may be ignored if the angle £ in Fig. 18 is suffi-
ciently small. This condition can be expressed by
OiOj' _ 2 L A<x> ^A
£••
OxC Rc + z0 xc
On the other hand d must not be too large, because
flat maxima and minima are undesirable. Thus a
compromise must be reached with respect to a suit-
able choice of d.
A qualitative consideration concerning the error
in determining c m from the spacing between two
maxima or minima of \Zt\ can be given as follows.
A change in frequency changes # according to eq.
(3.27). From Fig. 10 it is seen that an increase in
# corresponds to a clockwise movement of r around
the circle in the r-plane and consequently to a clock-
wise movement of Zyx in the Zu-plane. Moreover, an
i n c r e a s e in frequency changes the position of the
centre 0 / in Fig. 18 in the direction of the n e g a -
t i v e y-axis. Let 0 / in Fig. 19 correspond to the
maximum A' of | Zx | ; by a positive change Aa> the
centre 0 / may be shifted to 0 / ' corresponding to
the adjacent maximum A " of I Zt | . As may be seen
Increasing frequency

or Af< (5.3)
4>TIL

where, according to eqs. (3.21), (3.22), (3.27)

and (3.34)

YqEQclt l + #2 Decreasing frequency

0 for the bar,
4AH2 1~R*
ZQXC = (5.4) Fig. 19. Location of adjacent maxima A' and A" or A*
VqDQcl* l+R* and A** of Zt corresponding to a departure
for the l a t e Aco from the reference frequency (Ja> = 0).
° ~TXW r^R2 p '
and R= \r\e~2a™d.
from Fig. 19, the corresponding change in # , lead-
For example, the necessary change Af to describe a ing from A' in a c l o c k w i s e direction to A", is
full circle of Z\\, corresponding to Aft = 2n, is l e s s than 2 n (since # = 2 3t would correspond to a
given from eq. (3.27) by full turn around the circle, leading back to A ' ) .
The change Af in frequency, which, according
(A& = 2n). (5.5) to eq. (3.27), is proportional to the change
in # , is therefore l e s s than it would be if the
To keep Af as small as possible, the path-length d centre 0 / did not move. In other words, the
should contain a large number of half-wavelengths. maxima A' and A " are spaced nearer together than
ACUSTICA
Vol. 7 (1957) F. E.BORGNIS: ACOUSTIC INTERFEROMETER 171

they would be if the crystal impedance Z l c == O / O
did not change. If Af is inserted into eq. (4.25), a
value of c m s m a 11 e r than the correct value results,
because formula (4.25) is based on the assumption
that the change in # , corresponding to adjacent
maxima (or minima) equals 2 Jt exactly.
As the frequency d e c r e a s e s from the refe-
rence-frequency (Aoo = 0), the centre 0 t * in Fig. 19
may shift to Ot**, the corresponding maxima of
| Zt | being A* and A**. A decrease in / corresponds
to a decrease in # , and the circle for Zu is describ-
ed c o u n t e r - c l o c k w i s e from A* towards A**.
Here too the change in & between A* and A** is
l e s s than 2 Jt, and the maxima or minima are spac-
ed more closely than they would be if & equaled
2jt.
Thus the value for cm found from formula (4.25)
is less than the correct value, when derived from the
maxima or minima on either side of the reference-
frequency. The variable-frequency interferometer
involves i n p r i n c i p l e an error of this nature
in the determination of c m ; the error might be
made negligible if the shift of the centre Ot could
be kept sufficiently small.

APPEN DICES
A. Numerical values for the constants of an From this we obtain
X-cut quartz crystal H2/i S „E 0.88 • 10
H2
1. The bar in lengthwise vibration qE 1 + T E} = 8.8 • 10 1 0 newton/m 2 .

Thus H2/sT qD, which occurs in the treatment of the

The numerical values needed for the bar are s 22>
plate, is small compared with unity. The values for
d12 and exT, as seen from eqs. (1.1) and (1.2).
qE and H are somewhat different for the plate and
The constants, converted to mks units, are taken
the bar, as stated already in Part I, Section 3, but
from W. G. CADY [5] 13 . The page references below
still of the same order of magnitude.
are from this book.
The relative difference in frequency between se-
sg = a* = 1.27 • 1 0 ~ n m 2 /newton (pp. 55 and
ries resonance and antiresonance for the free plate
137), in the fundamental mode (n = 0) follows from eq.
d12 = — dn = — 2.3 • 10 12 coulomb/newton
(2.19):
(p. 219) , w a - w s _ Aw _ 4 H2
ef = 4.0 • 1 0 " 1 1 farad/m (k± on p. 414) . = 3.5-10~3
(o„ OJ„ %' er qL
From these quantities we derive The value of coty>/2 at s e r i e s r e s o n a n c e
qE = i / , « = 7.9 • 10 1 0 newton/m 2 , of the free plate, neglecting crystal losses, follows
H = rf12/af2 = - 0.181 coulomb/m 2 , from eq ( 2 . 1 5 ) :
. o)al 2 H2
#2/£V=1.04-10-2. cot q> 2 = cot - — = — — s w-j:
' 2cc (2n + l) n e qD
Thus the term H2/sT qE can be regarded as small in
_ ( 2 n + l ) jt Ao)
comparison with unity, as repeatedly assumed. The
2 oiA
relative difference between the frequencies for anti-
resonance and series resonance for the free bar in and for the fundamental mode (n = 0)
the fundamental mode (n = 0) is given by jt Aco
cot <p/2 = = 5.5-Kr3
cq. (2.10) : 2 coa
W . - Oia Ao) _ 4 H2 The slight departure of cot<p/2 from zero corres-
T E ;4-10~3
~ r2 "e q ponds to the fact that the thickness of the free plate
w, a>a
at series resonance is slightly different from XJ2 in
the fundamental mode. Indeed, we have
2. The plate in thickness vibration
( 2 n + l ) Jt Aco
The constants needed for the plate are qE = cn, 2cc ~
H = en, and sts, as seen from eqs. (1.25), (1.26), and hence for n
(1.27), ( 1 . 2 8 ) :
Ao)
qE = c^ = 8 . 7 - 1 0 1 0 newton/m 2 (p. 137) , = 0.9965-^-,
H = en = 0.173 coulomb/m 2 (p. 219) ,
ef = 3.9 • 1 0 ~ n farad/m (*^ on p. 414) . Actually, as is well known, the frequency measured
experimentally for series resonance of a free plate
13
Numerical values are also given in [3], pp. 90 is somewhat higher than the theoretical value, owing
and 91 and [15], pp. 84 and 85. to the fact that the plate is of finite area and not infi-
 ACUSTICA
172 F.E.BORGNIS: ACOUSTIC INTERFEROMETER Vol. 7 (1957)

nitely extended as assumed by the theory. The slight
departure of I from A 8 /2, as found above, is there-
fore chiefly of theoretical interest.
B. Numerical examples of interferometer
impedances
As an illustration of the foregoing theoretical con-
siderations we offer two numerical examples, one
for water and one for air. In each case the ultra-
sonic source is an X-cut quartz plate.
1 . M e d i u m : Water, / 0 = 15Mc/s
The following data are used:
ec = 2 . 6 5 - 1 0 3 k g / m 3 ,
c c = 5.8 • 10 3 m / s ,
em = 1 . 0 - 1 0 3 k g / m 3 ,
c m = 1.5 • 1 0 3 m / s ,
m0 = QmcmJQcc(i = 0.1 (eq. ( 3 . 1 7 ) ) ,
,4 = 1 cm 2 = 1CT4 m 2 , | T | = 0.9,
2<xm = 0 . 1 1 c m - = l l m - 1 .
1 I ^ii | m i n and can therefore not be ignored for fre-
From appendix A 2 we have H = 0.173 coulomb/m 3
and qD = 8.8 • 10 1 0 newton/m 2 .
The thickness of the plate is Z = Ac/2 = c c / 2 / 0 =
1.9" 1 0 ~ 4 m , assuming the plate to be driven at its
fundamental antiresonant frequency (eq. ( 2 . 1 8 ) ) .
z0 is found from eq. ( 3 . 2 2 ) :
D
Z2
H = Vq Qc 2 AH2 = 92 000 ohms.
The wavelength in water is A
= 0.01 cm. The quantity a m A m = 0.00055 < 1 ; ac-
cording to eq. (4.1), m± can be neglected in com-
parison with m0 and the centre of the circle for Z\\ can
be assumed to be on the real axis. A change in path
length of A m /2, corresponding to a full turn around
the Zjj-circle causes a contraction of the radius R in
the r-plane by the factor e~*m*m = 0.99945; evi-
dently this contraction can be ignored (see Part IV,
Section l b ) .
We choose several different values for the path-
length d ; the corresponding R is found from eq.
(3.27). The location of the centre XQ and the radius
Q of the (Zu/zo) -circle follows from eq. ( 3 . 3 4 ) ;
yc = 0 , since m1 = 0 and cot (p/2 = 0 . Maxima and
minima of | Zyx | are then given by z0 (XQ + Q) or
by eq. (4.10). Table I shows the results.
It is seen that the crystal is loaded considerably
by the water-column even at the minimal values of
| Z1{ | , as has been remarked at the end of Part IV,
Section 5. The resistance Zi c of the free crystal may
be assumed to have a value between 100 and 1000
ohms; it is of the same order of magnitude as
quencies of this order of magnitude. Since the order
of magnitude of | Z\\ | is proportional to m0/f02
(eq. ( 3 . 2 2 ) ) , | Zn | extends to much larger values
when the frequency is lowered, so that the influence
of Z l c may become negligible at lower frequencies.
2. M e d i u m : Air at atmospheric pressure and
0°C,/=lMc/s
Using Qm = 1.3 kg/m 3 and cm = 331 m/s one finds
m0 ~ 0.28 • 1 0 - 4 . The reflection coefficient | r \ may
be computed according to a formula given by HERZ-
FELD [10]:

Table I
Water, /„ = 15 Mc/s, area = lcm 2 , X-cut quartz plate (l = Xc/2),
l1z llil lImax
d 2<*/V R %C Zo QZo 1 ll
Imax 1 *" Imin
l*«lmin
cm ohms ohms ohms ohms

1 200 0.81 22150 21670 43810 480 91.2

2 400 0.72 14500 13750 28250 750 37.7
5 1000 0.52 8010 6560 14570 1450 10.0
10 2000 0.30 5510 3030 8540 2480 3.4
00 00 0 4600 0 4600 4600 1.0

Table II
Air at atmospheric pressure, / 0 = 1 Mc/s, area = 1 cm2,
X-cut quartz plate (Z = Ac/2).
Ll
2d/A m 1 1 max
d R xcz» QZo 1*111 1 Z l i Imin
1 ll
Imax l*l»lmln
cm ohms ohms ohms ohms

1 . 60.6 0.64 717 651 1368 66 20.7

2 121.2 0.43 437 317 754 120 6.3
5 303.0 0.13 311 79 390 232 1.7
10 606.1 0.018 301 11 312 290 1.07
00 00 0 300 0 300 300 1.00
ACUSTICA
Vol.7 (1957) F. E. BORGNIS: ACOUSTIC INTERFEROMETER
| T | = 1 - 4 . 4 - 1 0 - 5 V 7 = 0.96.
The coefficient of absorption may be found from
[ 3 ] , Table 64, p. 5 4 0 :
am = 2 • f • 10~ 1 3 = 0.2 c m - 1 or 2 a m = 0.4 c m - 1 .
The wavelength in air is Xm = c m // 0 = 3.31 • 10~ 4 m
= 3.31 • 10~ 2 cm. The quantity ctm Xm = 0.0066 < 1;
thus m1 as well as the contraction of a circle in the
r-plane due to a change of ^ m /2 in d, can be ignor-
ed (Part IV, Section 1 ) . Table II gives the numeri-
cal values, computed in the same way as for water
in Table I.
Owing to the considerably smaller value of m0 in
comparison with the foregoing example for water
the crystal is much less loaded by the air than by
the water, although the frequency chosen in the case
of air is 15 times smaller. This latter fact increases
the ratio | ZVl |/| Z l c | by a factor ( 1 5 ) 2 = 225. When
the frequency is further lowered, jZii| rises to
higher values by the factor l / / 2 and the crystal be-
comes heavily loaded even by a gaseous medium.
Fig. 20. Impedance diagrams for A = 1 cm2, Rc =
500 ohms, (a) water at 15 Mc/s, (b) air at
1 Mc/s.
Fig. 20 illustrates the two cases for water and air.
In (b) the vector construction for Z1 = RC + Zi[ is
indicated. A similar construction would be used in
(a).
173
C. Note on the "free" (non-radiating) face
of the piezoelectric crystal
All the results of the interferometer theory, as
presented here, have been based upon the assump-
tion that the crystal radiates into the medium within
the interferometer path only, whereas the opposite
surface is „free", that is, non-radiating.
Actually the latter surface is usually bounded by
air, rather than by vacuum, and the foregoing nume-
rical example for air indicates that, especially at
lower frequencies, the crystal may be considerably
loaded at a surface bounded by a gaseous medium.
If the calculations in Part I are carried through
with a modified boundary condition at x = 0, cor-
responding to a load at x = 0 represented by an im-
pedance Z m ' (or zm'), the following expression for
the impedance Z\\ replaces eq. (3.20) 1 4 :
1 - 2 j cot
•^m "T Zn
Z\\ = ZQ (CI)
cot <P
/" ~~ j
•^ni "T Zm •"
At resonance (Z = Ac/2) the cotangent terms vanish
and it is seen that z m ' simply constitutes an addi-
tion to zm , the acoustic load of the interferometer
path (cf. eq. ( 3 . 2 0 ) ) . Consequently at resonance
we have to add an impedance Z\{ , due to the gase-
ous medium at the f r e e surface, to the impedance
Z\\ of the interferometer path considered before. If
a plane reflected wave on the free side of the crystal
can be avoided by the use of highly absorbing mate-
rial, or suitable configuration of the reflecting sur-
face, or by having the path sufficiently long, Z\\
equals m0z0/2 (eq. ( 3 . 2 9 ) ) . As seen from the pre-
vious example for air, Z\\ then equals 300 ohms at
/ = 1 Mc/s for an area A of 1 cm 2 . Z\{ increases
with l / / 2 , becoming 30 000 ohms at lOOkc/s. If
the radiation into the free side strikes some obstacle
and produces a reflected plane wave, the gaseous
path behaves like another fixed-path, interferometer
in series with the true interferometer, and Z\{ may
induce a considerable complex impedance in addi-
tion to Z u . Proper attention therefore should be
given to the "free" surface of the crystal. The effect
of Z\{ is minimized when a plane reflector parallel
14
It may be noted that eq. (CI) also describes the
effect of a piezoelectric plate which is inserted nor-
mally to the direction of propagation of a plane com-
pressional wave and which is loaded at its input by an
electric impedance ZAB • One only has to solve eq.
(C 1) for zm' and to replace Z\\ by the expression
[ Z A B / ( 1 — j co CZAB)] — Z I C (see Fig. 8). However, the
formula so obtained applies only to the case in which
the centre of gravity of the plate is considered to be at
rest.
 ACUST1CA
174 F. E.BORGNIS: ACOUSTIC INTERFEROMETER Vol. 7 (1957)

and close to the free crystal surface is located at a (co/c c ) — j a c = 0) VQclqE for the b a r , and
distance of ( 2 n+ 1) A/4 . a) V(>c/qD for the plate, eq. (1.11),
,111 (co/cm) -jam, eq. (3.24),
The author is i n d e b t e d to D r . W . G. CADY for | r | eJ® reflection coefficient, eq. (3.8),
m a n y stimulating discussions.
r
£T, 8
s
permittivity, eqs. (1.2), ( 1 . 2 6 ) ,
(Received July. 1st 1955.)
6 phase angle of the reflection coefficient,
eq. (3.8),
Symbols & 2 o)(d/cm) - 9 = 4n(d/Xm) - 0, eq. (3.27),
A electrode area, h
^in acoustic wave length in crystal and medium,
Lh-, t'm density of crystal and medium,
b cot q>/2 ,
Q, o radius of circles in the iv- and w-planes,
Ct,Ci parallel and series capacitance in the equi-
eqs. (3.34), (4.9),
valent circuits of Figs. 3 and 5, eqs. (1.23),
(1.43), (1.45),
Cb, C p = C in Fig. 6, specialized for b a r and plate,
eqs. (2.6), (2.17), V Zli Imin
•Mi linax
1/1 Fli U
\ I I li | n m
= ycl = a>(l/cc) - j a c / = 2 J t ( / / A c )
eq. (4.12),

-jo>{l,
Cc» Cm velocity of sound in crystal a n d medium,
angular frequency for series-resonance of
D electric displacement,
the free crystal, eqs. ( 2 . 4 ) , (2.15),
d length of interferpmeter path (Figs. 1 and 7 ) ,
angular frequency for antiresonance of the
dm interferometer path-length corresponding to
free crystal, eqs. (2.9), ( 2 . 1 8 ) ,
a maximum or minimum of the electric
= 2*/.
impedance Zn, eq. (4.21),
E electric field-strength, References
F dissipation constant, eq. (1.9), BECHMANN, R., Dickenschwingungen piezoelek-
[1
f frequency = eo/2 JI , trisch erregter Kristallplatten. Hochfrequenz-
H piezoelectric constant, eqs. (1.3), (1.4) and techn. u. Elektroakustik 5 6 [ 1 9 4 0 ] , 14.
(1.27), ( 1 . 2 8 ) ,
[2 BECHMANN, R., Einige Anwendungen der piezo-
I. current at crystal input, elektrischen Zustandsgleichungen. Arch, elektr.
M l 5 •L'P L in Fig. 6, specialized for b a r and plate, Ubertragung 9 [ 1 9 5 5 ] , 122.
eqs. (2.6), (2.17), a] BECHMANN, R., Uber Dickenschwingungen piezo-
/ length of crystal b a r (Fig. 2) and thickness
L2
elektrischer Kristallplatten. Arch, elektr. Uber-
of crystal plate (Fig. 4 ) , tragung 6 [ 1 9 5 2 ] , 3 6 1 .
m 7 m g m / 7 c g c = ni 0 + j m 1 , eq. (3.17),
[3 BERGMANN, L., Der Ultraschall; 6th ed. S. Hirzel
n 0 , 1 , 2, 3 , . . . , Verlag, Stuttgart 1954.
g, L£,gL„D elastic stiffness, eqs. (1.3), (1.4), ( 1 . 2 7 ) ,
[4 BORGNIS, F . E., On t h e theory of t h e fixed path
(1.30), acoustic interferometer. J. acoust. Soc. Amer. 2 4
<7c0 + j U> F = qE for a b a r or q° for a plate, [ 1 9 5 2 ] , 19.
eq. (1.9),
[5 CADY, W. G., Piezoelectricity. McGraw-Hill Book
qm 9mo + j oyP, eq. ( 3 . 3 ) , Co., New York 1946.
R in Fig. 6, specialized for b a r a n d plate,
[6 CADY, W. G., A theory of the crystal transducer
eqs. (2.5), (2.16), for plane waves. J. acoust. Soc. Amer. 2 1
Re loss-resistance of the crystal due to absorp- [1949], 65.
tion and mounting,
2a [7 CADY, W. G., Piezoelectric equations of state and
R | r | e ~ « A eq(3.27), their application to thickness-vibration transdu-
T R e - i * , eq. ( 3 . 2 6 ) , cers. J. acoust. Soc. Amer 2 2 [ 1 9 5 0 ] , 5 7 9 .
S,T strain and stress,
[8 FORSYTH, A. R., Theory of functions of a complex
t thickness of the crystal bar, Fig. 2, variable; 2 n d edition. University Press, Cambridge
V sinusoidal driving voltage at crystal input, ( E n g l a n d ) , p . 605, S e c t i o n 2 5 8 .
w x + jy = Zn/z0, eq. (3.32), [9 Fox, F . E., Ultrasonic interferometry for liquid
IV x + jy = z0Yu, eq. (4.8), media. P h y s . Rev. 5 2 [ 1 9 3 7 ] , 9 7 3 .
•*•<> VQ coordinates of the centre of a circle in the [10 Fox, F . E . a n d HUNTER, T . L., T h e ultrasonic in-
w-plane eq. (3.34), terferometer with resonant liquid columm. Proc.
x
c> Vc coordinates of the centre of a circle in the Inst. Radio Engrs. 3 6 [ 1 9 4 8 ] , 1500.
«>-plane, eq. ( 4 . 9 ) , [11 HERZFELD, K. F . , Reflection of sound. Phys. Rev.
Y l / Z electric admittance, 53 [1938], 899.
Z\,Z\ electric impedance, Figs. 3 , 5, 8, [12 HUBBARD, J. C , T h e acoustic resonator interfero-
mechanical impedance of the interferometer meter : 1 : T h e acoustic system and its equivalent
path, eqs. (1.14) a n d ( 3 . 1 ) , electric network; 2 : Ultrasonic velocity a n d ab-
Zu electric impedance of the free crystal, Fig. 3 , sorption in gases. Phys. Rev. 3 8 [ 1 9 3 1 ] , 1011
Zu electric impedance of the interferometer and 4 1 [ 1 9 3 2 ] , 5 2 3 .
path, Fig. 8, eq. ( 3 . 2 3 ) , [13 HUBBARD, J . C. a n d ZARTMAN, I . F . , A fixed p a t h
Zn P — j Q HUBBARD'S electric impedance of the acoustic interferometer for the study of matter.
interferometer path, eq. (4.5), Rev. sci. Instrum. 10 [ 1 9 3 9 ] , 3 8 2 .
zo see eq. (3.21) a n d (3.22), [14 KOSTEN, C. W., A new method for measuring sound
2m (0)/ycqc)Zm, eqs. ( 1 . 1 6 ) , ( 1 . 3 6 ) , absorption. Appl. Sci, Res. B 1 [ 1 9 4 7 ] , 3 5 .
3 C , <Xm amplitude absorption coefficient in space [15 MASON, W. P . , Piezoelectric crystals a n d their
for crystal a n d medium (in P a r t I I a is application to ultrasonics. Van Nostrand Co., New
written in place of a c ) , York 1950, p . 6 1 .