Sie sind auf Seite 1von 5

Theoretical Basis of the Bouguer-Beer law of Radiation Absorption

FREDERICK C. STRONG', VilZar+ovaCollege, Villanova, Pa. Presentations of the m a t h e m a t h l law governing the absorption of radiation as a function of the thickness and concentration of absorbing material are customarily based upon experimental observation. A complete derivation is presented, based only upon modern theories of matter and energy. Considering the radiant power of a beam as the number of photons passing per unit time and the process of absorption as the capture of photons by molecules or ions, the probability of capture is related to a ratio of areas in an element of material. This is then integrated over three dimensions for identical photons (monochromatic radiation) to yield the familiar absorption law. The general case of a beam with a continuous distribution of wave lengths is obtained by further integration and the result is given in several forms. In the belief that some of the equations are difficult to comprehend, graphical interpretations are presented, followed by analysis of possible methods of employing the law in practice. Finally, cauges of apparent deviation are considered. Both the derivation and the graphical interpretations of the absorption law should make it more intelligible and fruitful for analysts.

HE author has observed that the relationship of absorption of radiation to thickness of absorbing material and concentration of an absorbing component is customarily presented as the result of experiment. For example, one reads that:
As the initial radiant power varies, the fraction absorbed by a given sample of material remains constant. The rate of decrease of radiant power with thickness and with concentration is proportional to the initial power:

cis its velocity, and Xis its wave length. ilbsorption is then to be interpreted as the capture of a photon by a molecule or portion of a molecule, thereby increasing its energy. Likewise, the radiant power is interpreted as the number of photons passing per unit time. Consider a parallel radiation beam (Figure 1) in which the photons are identical (radiation is monochromatic) and the cross section of the beam is S. Let the beam pass through a homogeneous absorbing medium with: (1) parallel faces perpendicular to the beam; ( 2 ) thickness bi (3) initial power of the beam entering the medium after reflection equal to P,,,; and (4)power after traveling distance x equal to Pm. Consider an element of absorbing material with cross section S and thickness dx (Figure 1). Passage of P , through thickness dx will change it by dP,. Because P, decreases by absorption, dP, is negative. Let dS, be the projected area of regions in the element which will capture photons. I t will be infinitesimal because the thickness is infinitesimal.

where dP, and dP, are the rates of change of power of monodx dc chromatic radiation, P,, with thickness and concentration, respectively, b is a fixed thickness, and a: and a; are positive constants, The usual presentation then continues with the integration of these equations, yielding

Pm Pm,o

= e-a'bc


or related forms, P,,, being the initial radiant power. [Radiant power P, rather than intensity, I, has been recommended by the Committee on Letter Symbols of the American Physical Society (6) and by the Joint Committee on Komenclature in Applied Spectroscopy, established by ASTM Committee E-2 on Emission Spectroscopy and the Society for Applied Spectroscopy.]
The author has often wished to see a derivation from theory only. Such a derivation can be carried out much more readily today, with the improved modern theories of matter and energy, than was possible in the times of Bouguer and Beer. The exponential nature of the law suggests that probability plays a part. This has been recognized, but apparently never followed up by a published derivation. For example, Lange (6) stated in 1932 that the exponential dependence of radiation absorption by sols could be obtained by a consideration of probability. Fundamentally the process of absorption by colloidal particles and by molecules and ions would be identical.

If a photon is passed perpendicularly through the element, the probability of its capture &-illbe the ratio of capture area to total area, dS,/S. If this is done at random a great many times (as would be the case for the photons in a beam), the count of the
fraction captured would be

- dPm and P,

would be statistically

equal to the probability of capture: dP, -P,




0 dx X b Figure 1. Diagram for Derivation of Absorption Law

Although it is true that any given beam will have a fixed density of radiation determined by the optical cystem and will, therefore, not correspond to random passage of photons, the probability of photon capture will be unaffected, as the material is assumed to be macroscopically homogeneous,

The quantum theory of radiation accounts for t,he phenomenon of absorption more readily than does the wave theory. According to the former, radiation consists of particles (quanta or photons) of energy hv = hc/X, where h is Planck's constant, Y is the frequency of the radiation (when considered as a wave),

Present address, Stevens Institute of Technology, Hoboken, N. J.


V O L U M E 2 4 , NO. 2, F E B R U A R Y 1 9 5 2
The material will in general contain more than one absorbing substance, each contributing to the area, dS,:
dS, = a:dn,

proached in practice but probably never quite attained. Filter photometers pass a wide wave-length band and spectrophotometers a band whose narrowness is limited by the intensity of the source of radiation, the sensitivity of the detector, and the resolution of the optical system (Figure 2) ( 4 ) . Even an isobted line has been shown to have width. Hence, it is desirable to obtain a form of the law which applies to a band of finite width. In Figure 2, it is apparent that the radiant power of the band is the area under the distribution curve-that is, the definite integral: P =L $ A where y is the vertical axis in the figure. (It is of course necessary to avoid integration over a path containing the points of discontinuity of curves a and b. However, this can be done by dividing the area vertically a t these points and integrating over each portion separately.) The meaning of y can be readily ascertained, as the integral is the sum of an infinite number of infinitesimal areas, dP: dP = yclX (20)

+ aidns + . . . . .


where a ; is the capture cross section of a unit quantity (molecule, mole, gram, etc.) of species I, dnl the number of units of species 1 in the volume element Sdx, etc. Substituting and changing signs, dP, _ p,


+ aLdn, + . . . . .


[As the question of whether a photon can be absorbed by a molecule depends upon whether the energy of the photon (hc/A) is capable of raising the energy of the molecule to a permitted state, the as will be functions of wave length.] Summing over the interval 0 2 r 2 b,


Substituting the spectrophotometric quantities, transmittance P ( T = ,), and absorbance ( A = -log T ) , and retaining the subscript m to denote the monochromatic case, An equivalent mathematical notation for this derivative would be P i and it is frequently so represented, though usually with the prime omitted.



= a,!

log e

= 0.4343 a

Equations 8, 9, and 10 may be considered as equivalent statements of the absorption law in its most fundamental formnamely, in terms of quantity of substance in the path of the beam. [According to a recent intensive study ( 7 )of Beers original paper (f),this concept of the basis of absorption was also that of Beer. However, he chose to express it in terms of length and compensated for the use of different concentrations by considering that he had in effect changed the length of path. In other words, he did not use concentration as an explicit variable, contrary to prevailing opinion. ] In the interests of practicality, both concentration and path length are introduced as follow: The volume, T, of the path through the medium is bS. Therefore:
1 b

Figure 2.
a .

Wave-Length Distribution of Bands

6. Exit slit wider or narrower than entrance slit

Filtered band

Exit slit of approximately s a m e width as entrance slit

In order to apply the absorption law for monochromatic radiation to a finite band, an infinitesimal interval of the band can be considered. For band width dX, the radiation is ( a ) essentially monochromatic and ( b ) of infinitesimal power, dP,. If dP, is reduced exponentially to dP by passage through a given solution,



which provides an improved definition of T , and a new statement of the absorption law for the monochromatic case:

since nl/V = cl, the concentration of component 1. Substituting,


p , = P , , ~e - a:bct - aibcz - . . .


P, In - = - a:bcl- aibc2 Pm.0

= =

.. . ,.

To obtain the expression for finite band width,

log P, PW

-ahl -a h

- .. ...


+ azbcp + . . . . .

e - a l b a - anbcz .. .d A


y~ -m

- alba




The above equivalent expressions for the absorption law involve monochromatic radiation, a condition that may be ap-

This is the correct form of the absorption law for finite band width. In order to evaluate the integral mathematically, the variation of PA.^, a;, a;, etc., with ?.must be known.

For application to analysis, the absorption law takes the form:

simplicity, it is assumed that az = a3 = . . . . . . . = 0, b is constant, and the subscript 1 is dropped from al and cl. The exponential decrease of P A with concentration of one component for constant thickness is shown in Figure 3. For both concentration and wave-length variable, Figures 2, c, and 3 can be combined perpendicularly to give the three-dimensional graph of Figure 4.

However, it is more common to measure rclative transmittance, Tr = P / P B , where PB is the radiant power transmitted by a blank containing all the substances present in the sample solution except the substance or substances being determined and all of the same thickness. If only one substance (component 1 ) is ))ring determined, and if the solution is sufficiently dilute, or the absorptivities of the other components are sufficiently small that their absorbance in the sample solution is equivalent to the absorbance of the blank, d p B = p x , ,e&


- a:hs

. .dX





Figure 4.

Surface of Equation Px
a = f(k) and b constant

P x . ~ ~ O - ~ ~ ~

The equation for P A = f(c, A) is that of a surface which interwith wave sects the PA,X plane in the distribution curve for length and the parallel plane through c = cz in the curve for P i = j ( c , , A). The cross-hatched areas are, from left to right,

Figure 3.

C Curve of Equation Px = PxJWak

b mustant


(= l > x


Dividing Equation 24 by Equat'ion 27,

which is another form of the absorption equation for monochromatic radiation. dP =

EB- a:ba e ah





and their ratio is T . The portion of the surface marked L represents a region of the band spectrum not absorbed by the solution. The form of a graph for comparison with a blank-i.e., P i = P ~ , IB - abc-will be the same. O The three-dimensional graph of T , = is shown in X Figure 5 . The intersection of the surface with the T,,&, plane will be a straight line at T , = 1 because here dP = dP,. The region L of Figure 4 becomes a horizontal rectangle. The areas corresponding to those evaluated in the previous figures no longer have significance, as T , cannot be integrated directly to give 7'. Figure 5 is transformed into Figure 6 by taking the negative logarithm of the vertical axis: -log T, = abc. The surface passes through the wave-length axis because for c = 0, A , = 0. The plane in which c = 1 intersects the surface in a curve which is the variation of a with X or the familiar absorption spectrum curve. Every plane perpendicular to the wave-length axis at bonie value X4 will intersect the surface in a straight line mith equation A , = a(X,)bc = kc.


Because the practical use of the absorption law in cheniical analysis involves integration of unknown functions, any one of a number of procedures can be followed.
-1. An experimental curve for the variation of absorbance with concentration can be determined, using a given band of fixed width and solutions of known concentrations. This curve can then be used for analyses under these conditions and with the same instrument. Even installation of a new radiation source might require revision of the curve.


The meaning of the foregoing equations may not be readily grasped. Figures 3 to 6 will perhaps make them clearer. For

V O L U M E 2 4 , N O , 2, F E B R U A R Y 1 9 5 2
13. The approximation that the absorptivities are constant in the wavelength interval can be made:

grands of the expression for the relative transmittance and the value of the absorptivity a t 8 given wave length calculated for known concentrations. This can then be used with any instrument for determining unknown concentrations, provided that the constants in the approximate functions are also determined for the instrument. Eberhardt (5) proposed appi oxiriiate functions for spectrohas photometry with the Beckman DU and similar instruments. Replacing a portion of his notation by that used in this paper, his function for P i is:

- p, e - 5; bci - 5; bcz - . . .


or^=- ~ , - - n ; b c i - a a ; b c 2 P


(37) The constants OL and p correct for slight deviation from the triangular shape of PA,^ caused by nonlinear dispersion. Xo = l/p(A1 XZ) and AA = l/a(X2 - XI).

Whether this approximation is sufficiently accurate can be determined experimentally by the procedure desoribed in A, a straight line indicating that it is. It will be observed that the approximation consists of applying the equation for monochromatic radiation,


to finite band width. When the sample solution is compared with a blank, the approximate equation becomes T7 and

- e--obc

= 1O-abc


A, = abc

where the subscript r denotes measurelnent relative to the blank.


1 Figure 6. Surface of Equation A , = abc


f(X) and b constant

For the exponential factor i n the function for the blank he suggests e - a:bc, - a i % - . . . 5X r(X) = r(Aoj/l

+ p(X



where r(A,) is a constant. This is satislactory if absorption is not great and the band not too wide. For the absorptivity function he substitutes one of t,wo alternative functions:
a ; ( h ) = a;(x,,

+ y(X - A,) + 6(A


, I

a:(A) = a:(A,) - yl(X - A,)

for ho - aAX 5 h for Xo 5 X 5 Xo

a:@) = a:(x,,

+ -&- A,)


5 A,

(431 (44)


Figure 5. Surface o Equation T, = f

a = f i and b constant ()

The monochromatic equation will give the best results when ( a ) the pass band is as narrow as possible, ( b ) the band is centered on a wave length for which the absorptivity is a maximum or a minimum, and (c) the concentration range is kept small. Condition ( b ) is desirable for two reasons. Variation of absorptivity with wave length is small and the distribution of radiant power with wave length (when triangular) is a t a maximum in the center of the band where variation of absorptivity with wave length is least.

These approximate functions are substituted in the equation for Y7 (Equation 32) and integrated. Eberhardt gives equations for the two values of T , correspondiiig to the function chosen for a; and also the simpler equations for cases where a = 0 = 1 and p = 0. Philpotts, Thain, and Smith (8)have proposed an error function as the approximation for n;(x) when x single absorption band in t,he infrared can be isolated:

C. Provided that the resulting curve is approximately linear and the instrument is suitable, transmittance or absorbance can be measured at varying band widths and extrapolated to aero band width. D. Approximate functions can be substituted in the inte-

where 1 is a slightly different band width measurement (distance between points of inflection). The exponential is expanded and e-a;(X)bc is again expanded. The integrals are evaluated by substituting five terms of the second expansion and assuming that is constant over the interval. From the resulting equation, a( ho) is determined. Hardy and Young ( 3 )have inverted the integrals by means of a Fourier transformation and a Taylor expansion, obtaining an

expression made up of a sum of derivatives. The treatment is very general, though mathematically complex, and deserves attention.

different temperatures] owing to changes in solvation. As this involves a change in the structure of the substance being determined, deviation is to be expected. Change in absorptivity with temperature can be expected for another reason: The distribution of the energy states of the molecules would be changed, altering their average capture cross section for photons of given energy. Photochemical decomposition of the sample may occur during irradiation, causing a change in concentration. The low intensity of pass bands of photoelectric instruments usually avoids this difficulty whenever it is theoretically possible. .4t high values of radiant power, a saturation effect might be observed. With a relatively large number of molecules in a higher energy state due to absorption, the absorptivity would be altered. The obvious remedy for such a hypothetical situation would be to revert to low power radiation.

E. An attachment could perhaps be devised for a spectrophotometer which would automatically correct transmittance measurements to zero slit width. Hardy and Young have indicated how this ideal operation might be carried out electronically.

Upon examination, it is realized that few deviations from the absorption law observed experimentally are true deviations, as they result from nonfulfillment of the conditions of application of the law. A number of possible situations are enumerated below. The phrase I . . .does not obey Beers law, , , commonly refers to the nonlinearity of the variation of absorbance with concentration. I the deviation is large, it usually indicates a mobile f equilibrium that is shifted by dilution. Obviously, the deviation is not a failure of the law but of the esperinienter in not including the absorption and correct concentrations of all components. Nonlinearity in absorbance is to be expected for a pass band finite width. Here the deviation is present not in the measurement but in the approximate equation employed. Scattering of a portion of the radiation represents a decrease in radiant power not caused by absorption, since some of it will not reach the detector. Fortunately] the relative error is usually small if absorption is significant. Fluorescence and the Raman effect represent emission that can cause a positive error in the measurement of transmittance. These effects are also usually small in absorptiometry. Absorptivities may change in concentrated solutions or at

(1) Beer, A., Ann. Physik (Poggendorf), 86, 78 (1852). (2) Eberhardt, W. H., J . Optical SOC.Am., 40, 172 (1950). (3) Hardy. A. C., and Young, F. M.,Ibid., 39,265 (1949). (4) Hogness, T. R.. Tscheile, F. P., and Sidwell, A. E., J . Phys. Chem., 41, 394 (1937). (5) Hughes, H. K., et al., Am. J . Phys., 16, 176 (1948). (6) Lange, B.. Z . physik. Chem., A159,288 (1932). (7) Pfeiffer, H. G., and Liebhafsky, H. A., J . Chem. Education, 28, 123 (1951). (8) Philpotts, A. R., Thain, W., and Smith, P. G., ANAL.CHEM., 23. 268 (1951).
RECEIVED April 17, 1981. Presented in part a t Pittsburgh Conference on Bnalytical Chemistry and Applied Spectroscopy, Pittsburgh, Pa., March
7, 1951.

General Method of Improving Precision in Multicomponent Spectrophotometry and Some Practical Consequences Thereof

DAVID FIRESTONEl, Polytechnic Institute of Brooklyn, Brooklyn 2, N. Y .

The principles of relative absorbance colorimetry have been extended to multicomponent systems. In order to achieve the highest possible precision in the concentration measurement, the analyst must work with absorptivity ratios which are in turn relatively determined. The principles of the method for obtaining these ratios and for conducting such analyses are developed herein. Application is made to a two-component system consisting of the permanganate and dichromate ions. A tenfold gain is obtained in the precision of the analytical results obtained by the relative method as compared with the results for this same system analyzed by the absolute method. Extension of this relative approach to multicomponent systems is outlined, as well as a general method for chemical analysis without sample weighing.

HIS paper presents a number of considerations dealing with the problem of analyzing multicomponent systems using spectrophotometric techniques, By a combination of a relative absorbance technique coupled with the use of absorptivity ratios precisely determined it is possible to reduce the error of measurement in cases such as these so as t o make their spectrophotometric determination comparable in precision to those of volumetric or gravimetric techniques. The principles of relative absorptiometry have been extensively treated in a group of papers dealing with the fundamental theory
1 Present address, Food and Drug Administration, 201 Varick St., New York. N.Y .

(7-9) as well as with a number of practical applications thereof (8, 3, 5, I S ) . In all of these papers the emphasis has been on the determination of the unknown concentration of an absorber relative t o that of a reference standard. Since such relative concentrations may be determined precisely, it follows that relative absorptivities may be determined with similar precision and, consequently, multicomponent systems possessing considerable overlap of their absorption bands may be analyzed with the same increase in precision available to systems already described in the literature. Although some other rrorkers have utilized relative absorp tivity ratios in multicomponent analysis, the essential character-