Please check your set up correctly. Particularly the opening/aperture/size of the detector. It may stop some portion of the light incident on it. If you have small aperture detector you can manage by optimizing the distance between the sample and detector, if you have any idea about the refracting properties of the sample. Also check the nature of the lens so that beam remains Gaussian in the whole path of the beam. Wish you good luck.

Thank you Pathik for your answer. The active area of the detector is high enough to get all the light. Assuming the portion of light does not go to the detector, why it should effect on the symmetry of the open-aperture curve? The curve should be symmetrical relative to the beam waste.

The nonlinear refraction is negligible, the results of the closed -aperture z-scan revealed that. Why and how the nonlinear refraction can effect on the open-aperture z-scan curve?

I recommend to check up attentively:

1. Whether really the initial beam has Guassian distribution.

2. Whether there is no diffraction because of a deviation of an initial Guassian beam from a propagation axis.

3. Whether there are no in the gone beam aberrations, for example, phase distortions of some order. Even at not coherent case (there is no a "point" photodetector, i.e. with small aperture) such distortions are possible. For example, the phenomenon like a coma may exist even for a case of a "point" photodetector (see e.g. Fig. 8 in our paper "Influence of phase distortions on the response of an optical heterodyne microscope", Laser Physics, 1996, V. 6, No. 4, pp.753-758).

Good luck

The light beam may not be normally incident on the sample. 

Thank you Alexandr for the answer, among listed by you reasons maybe the third one.

I suggest you to perform the OA z-scan exp. with variable energies.

Thank you Pathik for suggestion, unfortunately result is similar.

By induced non-linear effect such as self focusing or fanning at the laser waist, the scan can be asymmetric.

Also if your laser repetition rate it too high, you can induce thermal effect in the solvent that can be misinterpreted as non-linear origin.

Pierre-Alexandre Blanche can you say how self focusing, i.e. nonlinear refraction can effect on the symmetry of transmittance curve in the open-aperture z-scan? In the open-aperture measurements all the light is collected even that which is nonlinear refrected along the propagation axis; the curve should be symmetrical even in the presence of the self focusing, self defocusing, or no?

The repetition rate is 1Hz, and it is enough to induce heating of the nanocarbons in a solvent and further heat transfer to a solvent. 

Nonlinear effect see for exemple: http://www.scielo.br/img/revistas/bjp/v33n4/a35fig04.gif

1HZ should be low enough to avoid  thermal effect

Z-scan measures nonlinear response due to all possible effects. When you use long pulses (nanosec qualifies as long pulses) you may have thermal contribution, or some orientational NL due to clustering or so, or large nonlinear absorption may cause asymmetries too (it suppress the peak and enhances the valley) thick samples also are common for non symmetric curves.

The symmetric curve will be present when pure electronic refractive nonlinearities are in effect on thin samples, otherwise asymmetries are most likely present.

For CW we have a simple model to take into account different NL effects (OptExp 15 5 (2007) 2517-2529). Van Stryland has a nice recollection of the effects of different NL on the Z-scan technique for pulses

Sorry wrong button. For OA Z-scan the same holds. You'll need pure moderated electronic NL absoption for symmetric curve, thermal (nsec produces thermal) also poduces NL absoption assymetry (lensing) or if your sample is too long compared with the absorption lenght, orientational ,,,...

To be sure go to a short pulse source (femto) to obtain pure electronic NL absoption and from there

This http://www.scielo.br/img/revistas/bjp/v33n4/a35fig04.gif is typical closed or closed/open aperture z-scan, representing the negative third-order nonlinear refraction, not open-aperture curve.

Thank you C.G Treviño-Palacios for your answer. Can you say in what cases thermal mechanism produces NL absorption asymmetry? Should it always induce asymmetry or maybe there are some cases when not? I am asking because in a lot of papers, in particular paper of W. Blau et al., Appl. Phys. B 91, 521–524 (2008) the nonlinear absorption is caused by the nonlinear scattering on the thermally induced solvent and/or carbon vapour bubbles however the open-aperture normalized transmittance is symmetric. The paper is attached.

First, check the data with scanning in the opposite direction. If you still see the asymmetry again, it must be from the sample. It depends on lots of factors: sample form, thickness, homogeneity, pulse duration, wavelength etc..

Thank you Venugopal Rao Soma. The results of scanning in the opposite direction remain the same.

it is hard to say which is the mechanism. It is  a trade off between NL length, NLabsorption length and thernal difussion. If you can not have shorter pulses go to shorter samples. The paper you attached uses 1 cm cell,so start from there, Good luck

Thank you for your response C.G Treviño-Palacios. I have checked with shorter pulses, fs laser pulses and got a symmetric curve in the OA Z-scan measurements, so no any assymetry have been found. 

Good, so there is some drift going to nanosec; good luck finding out why