It could be that the harmonic peaks are higher order diffraction of the 5 32 nm light that is from the source that is being collected by the spectrometer I agree with Sergii that the geometry can help reduce the scattered light being collected and that a longer wavelength band pass filter may help. However, you might want to try to use a 532 nm blocking filter. or a higher optical density filter. Since you PL signals are small, even a small amount of the 532 going into higher orders of the spectrometer could be significant. But if you don't see a 532 nm signal through the spectrometer it could be the following.

If you are generating the 532 nm with a doubling crystal, or if your laser is doubling 1064. Your 532 nm long pass filter could be doing its job, but you may have some 1064 nm light going through the filter and then you are sing the high order reflections of that. You could separate that out before you excite your sample with a prism or dichroic filter etc, or block it with a filter and that could solve your problem.

I agree with the previous answers: You should check exactly where those lines are coming from. You can add a cleanup filter (essentially a narrow bandpass) after your laser in order to allow only 532nm light to go into your setup.

If it's 532nm laser light leaking through your detection filters, you can either add another 532nm notch (or longpass or whatever suits your needs).

HI, thank you all for your answers.  

I tried using a 532 nm notch filter.  It works better for as long as your PL signal is not on the multiples of 532nm.  As far as I understand, the notch filter is a stack of filters  with a periodicity that is reflected in the multiples of the wavelength you are blocking.  

Also, the blocking becomes wider the deeper you go in wavelength, that is:  At 532 it blocks 532± some nm, but at 1064 the range is much broader.  I have attached a PL signal from PbS QDs that whose PL is around 1060nm for demosntration.  That PL signal has both longpass filters: 532 and 850nm.

Even using a combination of 532nm longpass and 850nm longpass I get the strong harmonic peaks at 1064 and so forth.  I also attached a picture of how my setup is arranged.  The picture doesn't have the filters at the entrance of the spectrometer...just in case.

So far I'm still puzzled.   I'll try to realign completely the setup, try to get the PL signal to have as much normal incidence to the filter as possible.  If you have any other ideas, it would be much appreciated.

Dear Felipe Gerlein,

1. Do you use 532 nm from YAG: Nd pulse laser or CW laser?

2. If you use plulse laser for excitation, in order to detect PL only without 532 nm and its harmonics you should measurement PL with time reolution technique. This technique allows choosing time  to measure PL signal when or after excitation laser pulse off .  However, this technique requires : must using a digital oscilloscope device and the life time of luminescence signal must longer than pulse width of excitation laser (as I know, about 8 ns in catalog of YAG: Nd pulse laser and actually this time is delayed due to measurement devices system). In order to detect PL only with out 532 nm and its harmonics...,you can choose delay time for measurement (on digital oscilloscope), longer enough than pulse width of excitation laser. 

3. If you use CW laser for excitation, You can design a double-chopper (the sample in side two propellers of chopper), which allows fisrt propeller open when second propeller close and opposite. This means you alway detect PL signal without laser signal but the life time of PL signal long enough (about ms or slightly less).

Good luck

Thank you Thanh Nguyen Trong for your suggestion.  My laser source is a 532nm CW source.  It makes sense to have a double chopper to block the laser, I will take a look into that.  

Thank  you again everyone for your answers

Dear Felipe Gerlein,

 In case of CW source using double chopper, which is more simple than pulse laser source because the signal from PMT can be amplified by normal selective amplification technique as "lock in" (frequency and phase selectivity) while PL resolution time technique is more difficulities. Moreover, double chopper technique allows not only detection a pure spectrum of sample but also to avoid damage for PMT by laser radiative. About fabrication a double-chopper, to calculate with high accuracy for the gap (time) between the first and second propellers, the best of which is zero.

Aurelio,

The spectra at 1064 is indeed broad.  It is the photoluminescence signal from a batch of PbS quantum dots synthesized to emit at that wavelength.  That signal is not from the laser source.  In fact, the laser source is quite narrow and very selective around 532 nm