Has anybody tried the linearity for injection volume.i.e. 10,20,30,40,50µl injection of the same conc..Whether this will be linear or due to broad bands it will come out to be non-linear.And can we compare the results obtained by different injection volume?

We typically test for linearity by injecting 5µL through 50µL volumes, doing the 20µL injection 3 or 5 times along the way using a test mixture. We then check a given peak's linearity across the range of volumes injected and typically get at least 4 9s (r²>0.9999) from any well maintained autosampler. We also check the %RSD on the volume injected multiple times. We do this annually as part of our LC calibration check.

If you are using a column of "normal" dimensions, i.e. 15 cm x 4.6 mm or 3.9 mm i.d., and if you are not overloading the detector, the most likely cause is the solvent in which the sample is dissolved. If there is a mismatch between the mobile phase and the sample solvent, simply dissolve the sample in mobile phase. You can even use a solvent composition with a lower elution strength to make the injections.

If you get all of your sample onto the column, if you integrate areas, and, as mentioned, you are within the linear range of your detector you should observe a linear variation. Now, if you use an injection valve with a capillary loop you may loose sample due to laminar flow (see the Trouble Shooting brochure of Rheodyne, their web site, or Dolan, Snyder, Troubleshooting LC Systems, 1989, page 242). However, the 50% loss on filling a 20µL loop (Rheodyne ref.) could not be seen in our lab when a 75µL, or larger loop was filled. Apparently, laminar flow (or is this an absorption phenomenon?) breaks down in longer capillaries.