We have seen that clumps of dust with embedded young stars are needed to account for the FIR/submm SEDs of normal galaxies. We will refer to such clumps as "active clumps" (with embedded sources). Their main effect is to provide a warm component of emission to the FIR/submm SEDs.
An alternative concept for clumps is to consider them as passive molecular clouds with no embedded stars, such as those introduced by Witt & Gordon , or Bianchi et al. , and further explored by Witt & Gordon  and by Misiriotis & Bianchi . We will refer to such clumps as "passive clumps", since they are randomly distributed with no small scale (of a few pc) correspondence with the young stars. The effect of passive clumps on the FIR/submm SED is exactly the opposite of the effect of the active clumps, namely they provide cooler FIR SEDs. This is illustrated in Fig. 14, which shows results from a calculation made by Misselt et al.  for a spherical shell geometry with varying degrees of clumpiness. The reason why the grains are colder when the dust is distributed in passive clumps is simply that light is no longer reaching grains in the body of the clumps. Of course this effect only occurs when the clumps are optically thick. A medium of optically thin clumps would make no measurable difference to the FIR SED compared to the same dust homogeneously distributed.
Figure 14. The effect of "passive" clumps on the FIR/submm SEDs, taken from Misselt et al. . Model SEDs are for a range of density ratios k between the diffuse and clumpy media and for a filling factor ff = 0.1. All models are calculated assuming a spherical shell geometry. It is obvious that the effect of increasing the degree of clumpiness (decreasing k) is to make the SEDs cooler.
The effect of the clumps on the UV/optical/NIR SEDs is also dependent on whether the clumps are "active" or "passive". The effect of "active" clumps on the attenuation characteristics of spiral galaxies has been studied by Bianchi et al. , Misiriotis & Bianchi  and Tuffs et al. . Their effect has been also considered by Charlot & Fall  and, in the context of starburst galaxies, by Dopita et al. . Unlike the effect of diffuse dust in a disk galaxy, the attenuation produced by "active" clumps does not strongly depend on the inclination of the galaxy. Furthermore, the wavelength dependence of the attenuation is no longer primarily determined by the optical properties of the grains, but instead by the different blocking action of the clumps on the photons emitted by stars of different ages, and thus masses. This effect arises because longer lived stars are typically to be found at larger distances from their parent molecular clouds. Furthermore, the introduction of dust in the form of active clumps will increase the global attenuation of a galaxy more than if the same dust had been added to the diffuse interstellar medium. This can be physically attributed to the placing of the clumps in the direct vicinity of the UV emitting stars.
The effect of "passive" clumps on the attenuation characteristics of spiral galaxies has been studied by Bianchi et al. , Kuchinski et al. , Misiriotis & Bianchi  and Pierini et al. . Their main effect is to lower the global attenuation of the galaxy and only to a lesser degree to change the dependence of attenuation on wavelength. Details of the dependence of the attenuation on inclination and wavelength appear to depend strongly on the parameterisation of the clumps, for example whether the clumps have a fixed density or have a fixed density contrast to the diffuse medium (see Bianchi et al. ).
In summary, the major effects of clumps on SEDs are as follows:
In practice, the description of real galaxies requires the presence of both "passive" and "active" clumps. Indeed, the same physical interstellar cloud can act either as a passive or an active clump, depending on the observed wavelength. For example, molecular clouds will behave as "active" clumps in the UV, due to their strong spatial correlation with young stars, and as "passive" clumps in the optical/NIR, due to the smooth distribution of the old stellar population. This has an interesting consequence on the slope of the attenuation curve of spiral galaxies, namely that they will be steeper than predicted by homogeneous models. This may be important to bear in mind when considering the contribution of spiral galaxies to the Madau's plot. In the dust emission, clumps with embedded sources provide the short wavelength luminosity. Passive clumps are unimportant in terms of luminosity, but will increase the amplitude of the submm emission deep in the Rayleigh-Jeans regime.
Of course the results we have described are for a particularly simple situation, that of a two phase medium (diffuse and clumps). In reality the interstellar medium is turbulent, presenting a whole distribution of column densities along different lines of sight. The effect of this more realistic structure on attenuation has been considered by Fischera et al. , Fischera & Dopita , .