It is type IV nitrogen sorption isotherm. The slight shift in the desorption is normal. If you use solid-line instead of the open circles it will look better.

Hi Andrei, not sure I ever heard of such feature. So it would be very interesting but it requires serious checking about uncertainties... Large error on dV arise form small variation of temperature (even in controlled temp. room) and accuracy on P depends on calibration of P0  (using N2 or Kr, which law is used?) and sensors. Maybe some tests of repetability and inter-comparison would be interesting to confirm or deny this observation...

Dear Romain, we have Micromeritics Gemini VII 2390 V1.03 apparatus, and it registers saturation pressure as well as relative pressure (in example experiment it varied from 759 to 762 mmHg). Since the experiments performed at 77K, and the desorption curve stably goes under the adsorption curve, the temperature variation does not seem like an explanation in our case.

Dear Ali, I added the updated picture with lines instead of markers. The difference is small, but it occurs systematically. Still, may be it is just an artifact of pressure sensor or whatever in the apparatus.

Hey, may I make few proposals? First, I would support the Romain's suggestion and use some commercial standards, perhaps even several ones. Second, I think I have seen similar (even stronger) patterns in the data obtained by my colleague. The "rough" explanation was that the sample was not sufficiently degassed, and some impurities may remain on the sample; formally, one might suppose that impurities might be capable of demonstrating enhanced affinity as compared with a native surface becoming better exposed during desorption. The outcome would be that the initial portion of adsorption isotherm will exceed the initial portion of desorption isotherm, specifically, in the lower pressure range. Third, some tests might be elucidating: (1) a repeated cycle: sorption-desorption-sorption-desorption. Does this feature disappear in the second round? (2) longer degassing at (?) higher temperature, (3) increasing the sample mass; does it affect?

Thank you for updating the isotherm. I would not blame the instrument, yet!

Structural collapse of the sample and/or chemical interaction between the sample and N2 molecules should be taken into account as well. 

Thanks Mikhail for the complementary answer. Multiple ads-des step is very simple and seems highly relevant. Just a small coment for andrei on temperature. First, your sample is immerged in a dewar filled with N2 liq. This does not assure a constant temperature. The liquid is at equlibrium at its interface with gas, but there is a gradient of temperature within the depth of the dewar. This might change T during the measure when the dewar is slowly emptied. Secondly the temperature inside the device (not the sample) also matters because volumetric measurements are affected by variations of the room temperature. (dependency of molar volume with T). This usually induces biais during a first measurement (because device and pumps get warm) or because of daily variations.

I would just repeat the experiment but refilling the dewar near P/P0=1. I know that some Quantachrome instruments allow you to do this [only bulb in inside LN2]. It should make a difference if you are using Micromeritics [all cell is inside LN2].

As personal opinion, although fundamentally incorrect, the difference between adsorption/desorption is not large for any practical purpose.

I would suggest that go ahead and repeat the experiment. I have faced such type of errors during BET ASAP 2020  experiments. It may be due to improper degassing, not accurate purging and unreliable gas supply. Personally, I repeat the experiment and got better results.

Thanks

The error looks small between the two curves. As mentioned earlier try outgassing at the highest temperatura that your simple allows. If I am right in my thinking these instruments make a blank volumen calculation using helium. Some workers have found problems with this when the samples have narrow micropores as the helium does not come out of the simple if degassed while still in the dewar at 77K. So if you run the helium blank first then lower the dewar to pump out the helium. The problema is worse if you run the isotherm first and then the volumen calibration since nitrogen is even more strongly adsorbed in the narrow micropores. As a final point your isotherm is a Type IIb that is Type II in shape but showing a narrow hysteresis loop probably due to the formation of a meniscus between the finely divided particles at high relative pressure.

Have you repeated the experiment with those samples and with a fresh portion? Have you noticed any difference in measurement time between samples with the lower desorption points and normal behaviour? From my experience, in most cases, the reason for lower desorption values is a small leak in one fo the vacuum valves or the connection seal/gasket, where the problem increases with measurement time. Finally, not properly degassed samples could generate also such an effect.

it might be because of existance of micropores. Increasing dwelling time may help? Not too sure....