The expanded structure of foam transmits the energy of a crash much less efficiently than a solid member. At the area of impact, the foam readily conforms to the impacting object. This plastic  deformation absorbs the mechanical energy locally. The individual crushed pores undergo enormous amounts of local strain while the bulk of the foam can remain intact. This is the same principle of the controlled crushing of the engine cavity of a car in a collision. The sheets of metal bend and twist with a great deal of strain while the net movement of the outer surface of the car is relatively small.

Contrast this to the plastic flow necessary in a solid member. At the point of impact the solid undergoes compressive strain. If it were purely viscous behavior (like a gel) the impact would cause plastic flow without work hardening or shear frictional forces. The solid member initially indents plastically, but the elastic strain field grows rapidly, transmitting the deformation energy as an elastic wave. This wave can reflect back from a free surface or concentrate to result in ballistic fractures at corners or edges. In either case little of the crash energy is absorbed by the solid.

The expanded structure of foam transmits the energy of a crash much less efficiently than a solid member. At the area of impact, the foam readily conforms to the impacting object. This plastic  deformation absorbs the mechanical energy locally. The individual crushed pores undergo enormous amounts of local strain while the bulk of the foam can remain intact. This is the same principle of the controlled crushing of the engine cavity of a car in a collision. The sheets of metal bend and twist with a great deal of strain while the net movement of the outer surface of the car is relatively small.

Contrast this to the plastic flow necessary in a solid member. At the point of impact the solid undergoes compressive strain. If it were purely viscous behavior (like a gel) the impact would cause plastic flow without work hardening or shear frictional forces. The solid member initially indents plastically, but the elastic strain field grows rapidly, transmitting the deformation energy as an elastic wave. This wave can reflect back from a free surface or concentrate to result in ballistic fractures at corners or edges. In either case little of the crash energy is absorbed by the solid.

may be it comes from high porosity structure.. 

As Bob Jarret wrote metal foam thanks to high porosity absorb the energy and diffract it. Braking bridge in any porosite material need energy. This energy is higher than in solid materials, that why metal foam are used in car and in military use.      

Just small remark to Bob Jarret answer

Basically you van have foams from metallic alloy swhich have more brittle than ductile behaviour. Then the brittle fracture of broken foam layer absorbs the mechanical energy locally. The individual cracked pores undergo enormous amounts of local strain and crack propagation while the bulk of the foam can remain intact. Then the crushing of foam continues from this damaged layer to surroundings via crack propagation. The impact energy is used on continuous crushing of part of foam again and again.

Dear all,

just as an addition to the previous answers, to absorb energy in a crash, you need some strength level of course and some deformation path. Any conventional, two-phase metal foam (gas, metal) will have a much reduced strength - by rule of thumb, a quarter of strength at half the density -, but collapse of the cells in compression gives you some deformation path at roughly constant strength until you reach the densification region. For many crash applications, this lowered strength may be beneficial as it reduces the deceleration suffered by whatever you want to protect with your foam. If you put a block of solid material in place of the foam, naturally, the level of energy needed to deform it would be much higher, but still the material would not necessarily do the trick of protecting any object behind it, for the aforementioned reason. What you could of course do is replace the foam with some shaped metal that maybe matches the foam in terms of the amount of metal and gas: A hollow section like a tube, maybe. In this case, however,  you need to consider effects like buckling as a major mechanism of failure. In your force-displacement curve, you will see this as an initial peak, followed by a lower (in terms of force levels) region which may show additional peaks that reflect additional buckling/folding events. Foams usually do not show this behaviour, and thus have the potential for a force-discplacement curve that comes nearer to the "ideal" energy absorber, where the section used would be horizontal.

There is much to add, of course. I just want to make one additional comment to the answer by Jaroslav Kováčik: It is not necessarily brittle fracture that is at the core of the effect he describes. Local absorption if energy can also occur via (ductile) bending of cell walls, and actually, this can be influenced - for details, see the paper to which I provided a link below: Here we did heat treatment on different kind of aluminium foams, thus adjusting their strength and ductility. Interesting effect was, from my point of view, that very brittle and very ductile materials both deviate from the horizontal shape of the stress-strain curves often perceived as "ideal": In ductile materials, strain hardening leads to a positive inclination of the curve, while in brittle ones, there is a downward slope. In effect, the optimum may be the intermediate type of performance.

Hope this is of interest.

Best regards,

Dirk

http://www.sciencedirect.com/science/article/pii/S0921509302005828