I am working on vanadium redox flow batteries using carbon felt as electrode material and V(IV) as electrolyte. Whenever I increase loading of catalyst on carbon fiber, kinetics decrease.
The rate of the heterogeneous reaction is directly proportional to the specific surface area of the catalyst. If you have an optimal specific surface area, then the reaction rate is maximum. By adding more catalyst to the surface of the carrier (carbon fiber), you decrease the specific surface area of the catalyst. There should be not a lot of catalyst, but optimal for maximum reaction rate.
Data Yu. A. Mirgorod, R.V. Grebennikova. Glossary on Physical Chemistry
Yuri Mirgorod answer is great. Specific surface area is directionally proportional to the reaction rate. Over loading the catalyst can actually block reaction sites; think of two Ni catalysts, one as a piece of metal and the other nanoparticles. The catalytic activity of the nanoparticles will outperform the metal.
Other considerations are the flow rate of feed gasses or ions in the electrolyte, which can drastically effect the catalytic activity.
The answer by Yuri Mirgorod is excellent. I would only add (having worked on Pd/C and similar materials) that the active metal size tends to increase with increased metal loading (thus the reduction in specific surface area (SSA) as Yuri mentions). For example, the activity per unit mass of metal will be generally much higher for a 0.5% Pd/C than for a 5% Pd/C. The extent of fineness of the metal catalyst is called 'dispersion' and it's usual to talk about turnover frequency (TOF) in modern-day heterogeneous catalysis rather than reaction rate (as it was in my day). Poisoning of a catalyst by S or other materials also needs consideration and a more active catalyst (smaller particles) may poison more easily. Solutions such as alloying the active material can work in certain circumstances.
In addition to the main reason explained by Yuri Mirgorod and the other important points mentioned by Alan & Alec, I would like to add another possible reason related to the microstructure of the active component rather its texture. It is important to note that the possible increase in the size of the active component, due to the higher loading percent, affect the microstructure of the surface of the active component via decreasing the percent of the available corner sites and the edge sites which are usually more active than the face sites. Moreover, it changes the surface energy of the active component as well as the solubility of the impurities in it.