Using the Micromeritics instrument (or any other gas adsorption instrument), measure gas adsorption isotherm of N2 at 77 K on your sample, then perform pore size distriution analysis with the firmware software already installed on the instrument. If you can start from P/Po ~ 1e-5 and collect a large number of data points (~80-90) up to P/Po=1 you will get a good porese size distribution starting from pores ~ 1 nm large. The best option for analysis is the DFT method and the slit-shaped pore model *for carbons). For narrower pores (from ~ 0.35 nm to ~ 1 nm) you should use CO2 adsorption at 273 K and the same software for analysis of adsorption data.

Thanks for your reply. Is the number of data points in the low pressure region (P/Po less than 0.001) can affect the pore size distribution result. I had run a measurement with fixed dose of 2 cm3/g .and found that the liquid nitrogen in the container was finished before the measurement completed that causing abnormal shape of isotherm. I tried to refill the container during analysis with pausing the measurement ...and that too generate abnormal isotherm. Please tell me what should be the minimum number of data points (fixed dose amount) required for PSD.

If the material is highly microporous you should use a small size sample (< 10 mg) to allow the run to finish before you run out of L-N2. Second, rather than fixing the dose size, you better fix the pressure table and allow the machine to select the dose size accordingly. It's a matter of trial and error before you get to know the sample and the instrument response. I use Quantachrome Autosorb instrument. They have two pressure transducers (typically) but the one I have has an additional pressure transducer for max 1 Torr. This allows me to start from very low P/Po range (1 E-7, more often 1E-6). If you don't have the 1 Torr transducer, you probably can start from 1E-5 P/Po. Then, use one of the preinstalled pressure tables (e.g. "Micropore analysis" which is on the Quantachrome machine) and go with it. If such a table is not available for the Micromeritics instrument, you can build one yourself. Ask for 5 points per decade of P/Po and you should be fine.

Be very careful with weighing the sample, because using small amounts (~! 10 mg) might introduce large errors in weighing.

Minimize as much as possible the de dead volume in the measuring cell.

Thank you very much for your detailed reply.

Other important question is data processing after. Different models can be applied to obtain PSD, and the selection of one of them is very important to obtain accurate results.

Pore structure of carbon can be determined using Nitrogen absorption rate using the Brunauer-Emmett-Teller (BET), Langmuir equation and Dubinin-Radushkevish (D-R) analysis software to provide the adsorbed amounts of the adsorb or system pressure changes as a function of time.To evaluate the kinetic separation efficiency of nitrogen, the adsorption rates for nitrogen can be taken separately at 298 K and a constant starting pressure within 1 min. Finally, the overall kinetic separation efficiency can be calculated by dividing the adsorption capacity of nitrogen at approximately 1 min. The BET standard equation can be applied in the relative pressure range from 0.05 to 1. This equation graves a linear relationship between 1/VA [(P/Po)-1] and (P/Po), although the linearity is restricted to only a limited part of the volume adsorption isotherm].Then the micro pores size distribution of carbon sample can be estimated using the Horvath-Kawazoe (HK) method.

I am sorry to say that, but I disagree with most of the comments added by Mr. Abdelrahman.

(1) The pore structure is determined from equilibrium adsorption data, not from kinetic adsorption data, so there no need to measure adsorption rates or pressure changes as a funciton of time.

(2) I do not see the reason to determine kinetic separation eficiency of nitrogen (separation from what ?).

(3) The BET equation applies at relative pressures from about 0.05 ot about 0.3 (not 1). (4) The HK method is a poor alternative to the DFT method, which became a standard in recent years, since it is available in the software packages installed of most commercial adsorption instruments.

I completely agree with Prof. Contescu.

Concerning DFT models, you should know your type of carbon. Depending on type of pore or surface heterogeneity, DFT derived models (non local, NLDFT, or quenched solid, QSDFT) could better fit your adsorption isotherm.

As you suggest data processing is also very important after the N2 adsorption measurement. I have two models N2-DFT and N2-NLDFT and they are generating different PSD. The PSD derived from N2-DFT is also sensitive towards pressure range, I got two different PSD for pressure range upto 0.1 and upto 0.97. While PSD derived from NLDFT is same for both pressure range., Other selected parameters [like model type (DFT), Geometry (Slit) and regularization (None)] are same for both DFT and NLDFT data processing. The sample used was activated carbon.

Please suggest me which model should I use to derive PSD and the suitable regularization for that (None/Low/Med/High/entered).

Please find attached the PSD derived from original DFT and NLDFT models in two different pressure range.

NLDFT pressure range upto 0.99

DFT , pressure range upto 0.1

NLDFT pressure range 0.1

Best option to be sure that the model that you applied to model your isotherm is the appropriate one, is the goodness of fit. You can compare experimental isotherm with the modelled one. Special attention in the range of partial pressure corresponding to the knee. According to your files, it seems that best fit is obtained NLDFT 0.99.

I have the same opinion: it seems that NLDFT model up to 0.99 P/Po is the best choice. If you have these options, try also different geometries in the NLDFT model, such as slit or cylindrical pores, or combinations slit-cylindrical. More advanced is the QSDFT model, which assumes some roughness on the pore walls, a more realistic situation. There are indeed variations between the results of different models, and this has to be accepted because all calculations are model-dependent. Using a mild regularization level helps (and is recommended) because this would help "smooth" the sharp peaks and the zig-zag appearance in the low pore size range. But don't over-use it or you will loose details of differential distribution.

I am concerned about the first 10 or 15 data points, where apparently the equilibrium pressure first goes down with each more adsorption point and then returns to the normal increasing trend. This is not normal, pressure should change monotonically from low to high. The DFT procedure has probably eliminated these points from calculation. I suspect that the equilibration time of only 30 secoind is too low. Diffusion is slow at 77 K and you may not have true equilibrium points at very low pressures. I recommend equlibration times of 2 or 3 minutes.

Pore size distribution can be characterized by application of the Brunauer-Emmett-Teller (BET), Langmuir equation and Dubinin-Radushkevish (D-R) analysis software The BET standard equation you can applied in the relative pressure range from 0.05 to 1. This equation gave a linear relationship between 1/VA [(P/Po)-1] and (P/Po) although the linearity was restricted to only a limited part of the volume adsorption isotherm .Then the micropores size can be calculated using Horvath-Kawazoe (HK) method..

Last time i mansion kinetic separation efficiency of nitrogen for one min.,it is a technique any one want to run BET measurement should go through it, also he should do heating process to remove any gas taking place inside pore.

BET measurement is not enough we need other model to check. you can try Small angle scattering techniques by applying the Porod’s low which considered that when X-rays penetrate the entire sample volume, SAXS gives a measure of the surface area of all the pores, and derived a general expression for scattering in a two-component system (e.g., carbon and void) of an arbitrary structure called Porod’s law. This law considers that SAXS is caused by electron density fluctuation in the scattering medium on a scale of 0.5-200 nm; it is, therefore, applicable for micro-meso and macropore characterization of your carbon sample.

BET measurement is sufficient and there is also device in the case your sample contains a significant portion of micro pore change N2 adsorption by diluted Krypton adsorption (smaller molecules, I assume 1% Krypton).

Let's refocus: The question was "How to measure pore size distribtion of microporous carbons ?".

Measurements of BET surface area is not equivalent with measuring the pore size distribution. Moreover, the premises of BET calculation do not apply to microporous materials, because the mechanism of multilayer adsorption on random surface patches does not apply in micropores. For this reason, BET method for calculation of surface area from adsorption isotherms overestimates the results for microporous materials, However, the BET method is still used with these materials for the lack of a better method. It only gives a number (m^2/g) that is useful for comparative purposes for microporous materials (and others). But if the goal is to determine pore size distribution, then the DFT methods are regarded as the best choice available today. When using DFT (in any of its current variants, NLDFT or QSDFT) it is the researchers's task to select the geometry of pores (slit, cylindrical, or combinations of both, or others) in ther carbon. Moreover, the routines available for DFT calculationas are limited to a number of gas, material, and pore geometry combinations.

Dear All,

I had successfully determined micropore-size distribution by using DFT and nitrogen (at 77K) and CO2 (at 0C) adsorption isotherms. I would like to express my sincere gratitude and thanks to Dr. Contescu, and Dr. Maria Dr. Sanati, Dr. Abdelrahman for thier important suggestions and discussion.

Govind

Dear Govind,

Congratulations ! Glad to hear that !

Dear Govind, congratulation. It is great pleasure to hear that

Agree with Dr Cristian and Dr Maria.

But for microporous structure I thought BJH is better than DFT.

BHJ model is only valid for mesoporous materials

i agree with the opinion that BJH is good for mesoporous materials

van any one read my data for BET porosity?

I can help.

Dears,

I am not that experienced with the experimental and analytical procedures of gas adsorption so would you please tell me the exact relative pressure ratios for determining;

A- the micropore radii (using DFT or HK method)

B- the BET surface area (in case of micro and mesopores)

C- the micropore volume using t-plot and alpha-plot

D- the mesopore volume (is the 0.6-0.8 relative pressure using t-plot correct?)

E- the mesopore size distribution using BJH and DFT

Thanks in advance,

Ahmed

Ahmed,

You ask the right questions. However, the correct answer is: "it depends". Like with any other analytical techniques, you need to adapt your method and parameters to the properties of the sample you want to measure. A very good book that contains a lot of practical answers and theoretical back ground is "ADSORPTION" by F Rouquerol, J Rouquerol, K S W Sing, R Llewellyn, G, Maurin (Academic Press, 2014). If you are in a hurry and need quick answers my advise is to check the web site of Quantachrome. They have numerous technical notes that treat specific situations and will give you answers to your questions. This is much better, in my opinion, than looking after "one size (parameter) fits all". Another advice is this: Don't get discouraged. It may take a few unsuccessful trials (frequent in dealing with vacuum systems) but you will be on the right way.

Dear Dr. Contescu,

thanks a lot for your answer and suggestions. Really, I am not going to perform gas sorption measurements but I will talk about the gas adsorption in my defense (because I will compare adsorption data with positron annihilation data) and I got confused to define the the correct relative pressure range of each textural property ! Just I wanted to know approximated (or common) values , that is all.

Hi, thanks everybody for a fruitful discussion. Dear Cristian I Contescu and Maria Teresa Izquierdo ,Mr.Govind's files for NLDFT Standard Deviation of Fit: 1.47615, cm³/g STP and for DFT Standard Deviation of Fit: 0.21585, cm³/g STP. Still why would you suggest NLDFT over DFT ?

Thank You

Hi Vishnu Surendran ,

NLDFT, as well as QSDFT are both DFT schemes to obtain pore size distributions. As Cristian I Contescu mentioned above, the QSDFT method (as developed by Alex Neimark) is more advanced than the NLDFT (Lastoskie) due to the consideration of the roughness of the walls.

QSDFT is then more recommended so far (read the answers by Cristian carefully).

Govind Sethia did the analysis in Micromeritics software; as far as I know it does not have the QSDFT algorithms implemented (just NLDFT). QSDFT are currently implemented by Quantachrome.