1 Introduction

Gas–solid chromatography is an important hydrogen isotopic analysis method, which is based on relative vapor pressures and adsorption affinities of hydrogen isotopes for a stationary phase such as activated alumina (Moore and Ward 1958, 1960; Lässer et al. 1999, 2001; Kawamura et al. 2000). Activated alumina (γ-Al2O3) was first reported in 1958 to separate nuclear spin isomers of hydrogen isotopes (Moore and Ward 1958). To achieve baseline elution separation of H2, HD, D2, catalysts such as Fe2O3 (Moore and Ward 1960) and MnCl2 (Kawamura et al. 2000) were deposited on it to accelerate ortho-para inter-conversion and cause the isomers to appear as a single chromatographic peak. The elution nature of the modified alumina was assumed to be related to quantum sieving effect (Cai et al. 2013; Wang et al. 2011) which is caused by micropores (pore size less than 2 nm).

However, a zeolite with plenty of micropores, is not a good candidate for the elution due to relatively long retention time (Whisnant et al. 2011). The pore size distribution (PSD) according to the Barrette-Joynere-Halenda (BJH) method indicated the coexistence of micropores and mesopores (pore size at 2–50 nm) in the alumina (Cai et al. 2013; Wang et al. 2011). Hydrogen isotope affinity for micropores is proved to be different from that for mesopores (Chu et al. 2006). Until now it is unclear how the micropores or the mesopores individually influence the elution performance.

To better understand the elution nature of the alumina, it is essential to first know its textural properties. Although BJH is the most used macroscopic method for mesopore analyzing, it is not suitable for micropore assessment. Besides, it is found that for a size range between 2 and 10 nm, pore size is underestimated up to 25 % (Villarroel-Rocha et al. 2014). Instead of BJH, the microscopic computation approach density function theory (DFT), is proved to be a good tool to analyze both capillary condensation and micropore filling (Rouquerol et al. 2012).

As MnCl2 deposition content varies both surface area and pore volume of the alumina, and changes retention times of hydrogen isotopes (Cai et al. 2013; Wang et al. 2011). In this article, modified alumina samples with different deposited MnCl2 content were synthesized to obtain different textural properties and different elution behavior. The DFT assessment was applied to aid in establishing a more accurate relationship between textural information of the modified alumina and its elution performance.

2 Experiments

2.1 Materials

Carrier gas for the elution test is high purity Neon, which was purchased from Dalian Special Gases Company, Dalian, China, with a purity guaranteed as 99.999 %. Gas sample for the test was H2–D2–He mixture diluted with Neon (its purity >99.999 %), which was purchased from SWRDICI (The Southwest Research & Design Institute of the Chemical Industry, Chengdu, China). The nominal composition of H2, D2, He, is 4.97, 4.97, 4.75 %, respectively. At the presence of Pt catalyst, hydrogen exchange equilibrium between H2, D2 and HD is established at room temperature before the test.

Analytical grade activated alumina (80–100 mesh) and MnCl2·4H2O were purchased from Aladdin Chemical Incorporation. The starting alumina without modification was labeled as Mn00. Three kinds of modified alumina with different MnCl2 deposition content were synthesized and labeled as Mn01, Mn02, Mn03, the nominal molar ratio MnCl2:γ-Al2O3 = 1:10, 1:5, 1:2.5, respectively. A typical deposition procedure (Mn02 as example) was described as follows: 19.8 g MnCl2·4H2O were dissolved in deionized water to form 40 mL MnCl2 solution. 52.4 g alumina powders were mixed with the solution. Ultrasonic oscillation to the mixture was applied for 2 h. The mixture was then dehydrated in a far-infrared oven at 150 °C for 12 h.

The modified and unmodified samples were sieved and the sample powders with 80–100 mesh were collected for column package. Because elemental Mn content of the sieved powders determined by ICP-AES was 3.9, 7.6, 13.2 wt%, for Mn01, Mn02, Mn03, respectively, hence the actual amount of MnCl2 deposited was 8.9, 17.4, and 30.2 wt%, for Mn01, Mn02, and Mn03, respectively.

The powders were packed into a 3 m × 1/16 in. (O.D.) stainless steel tubing with aid of a vacuum pump. The internal diameter of the tubing is 1.2 mm. In order to prevent the particles being pumped away or blown out of the tubing, glass wool and stainless steel filter were added into one tube end before packing (Barry and Grob 2007), and added into the other end when the package was finished. The volume of packed samples is 3.1 mL.

2.2 Materials characterization

2.2.1 Nitrogen sorption isotherms

Nitrogen sorption (adsorption–desorption) measurements for the samples were performed at 78 K on a Micromeritics 3Flex volumetric adsorption analyzer. The surface area measurements are based on the BET method within a pressure range from p/p0 = 0.05–0.3 for the samples. BET surface area is calculated from the adsorption isotherm using the BET equation.

The pore volume is calculated from the amount of N2 adsorbed at a relative pressure of 0.99. For determination of pore diameter and pore size distribution, the majority of studies employ the BJH method which, however, is known to underestimate the pore size for pores smaller than 10 nm. A more reliable method is based on the density functional approach (DFT). Therefore, results from both methods are presented to demonstrate the difference. The analyses were made using the models provided by Micromeritics.

2.2.2 Hydrogen isotope elution experiments

The packed column was coiled with a diameter of 10 cm, which was installed in the chromatography instrument Agilent 6890 N. It was activated under the Ne carrier gas with a flow rate of 5 mL/min for 3 h, column temperature was maintained at 170 °C during the activation. The column was cooled down to room temperature after activation. Then the activated column was immersed in a liquid nitrogen dewar, and the flow rate was regulated to 10 mL/min. The instrument standby was ready for sample injection when the baseline noise is less than 1 μV. Sample loop volume is 50 μL, and injected pressure was controlled at 50 kPa. A thermal conductivity detector (TCD), which is suitable for packed columns and inorganic solutes, is used to detect separated solutes. Its working temperature is kept at 200 °C.

3 Results and discussion

3.1 N2 adsorption–desorption isotherms

Nitrogen sorption isotherms of the alumina and its modified samples were shown in Fig. 1. Shape of the four isotherms is similar, with an initial sharp uptake at a relative pressure below 0.05 followed by a more gradual adsorption. A hysteresis loop was formed between the sorption branches. According to the IUPAC classification, such isotherms can be sorted as mixed type I and type IV isotherms which are attributed to samples containing both micropores and mesopores (Schüth et al. 2002). Contrast to mesoporous molecular sieve such as SBA-15 (Rouquerol et al. 2012), which has sharp adsorption and desorption branches, the gradual branches of these samples are indicative of a wide mesoporous size distribution.

Fig. 1
figure 1

Nitrogen adsorption isotherms of the alumina and the modified samples

Full size image

The equivalent surface area of each sample (both micropores and mesopores are included), SBET, was calculated with the BET method and shown in Table 1. The mesoporous volumes, VBJH, which were calculated via the BJH method applied to the adsorption branch of the isotherm, were listed in Table 1. The presence of MnCl2 in the alumina matrix affects the BET surface area, and the pore volume, SBET decreases from 228.1 to 98.7 m2/g, and VBJH decreases from 0.483 to 0.219 m3/g. The more MnCl2 content is deposited, the more area and volume is lost. These phenomena may be explained by the fact that more deposited MnCl2 occupies more pores and blocks more channels.

Table 1 Some textural parameters of the alumina and the modified samples
Full size table

Pore size distribution of the pore volume, which was computed with the DFT model, was shown in Fig. 2. The dominated pore size was denoted as rDFT, which was listed in Table 1. The dominated pore size calculated by BJH model (denoted as rBJH) was also listed in Table 1.

Fig. 2
figure 2

PSD of pore volume of the alumina and the modified samples modeled by DFT

Full size image

The values of rBJH in this experiment differ from those in the literature, such as rBJH = 1.7 nm at microporous range (Wang et al. 2011), or rBJH = 7.3 nm at mesoporous range (Cai et al. 2013). These differences may be related to different synthesis process of acitvated alumina. Due to rBJH < 10 nm, BJH method based on capillary condensation will overvalue this effect and related vapor pressure, consequently underestimate values of pore size. In this experiment, rDFT is about 5 nm larger than rBJH, which is a good demonstration to show the limitation of the BJH model. As shown in Fig. 2, 0.7 nm micropores existed in the starting alumina, and disappeared after MnCl2 deposition. The volume of 1.5 nm pores was slightly decreased with increasing MnCl2 content. The mesoporous volume with wide pore size range, decreases as the MnCl2 content increases, this trend is in accordance with that in the literatures (Cai et al. 2013; Wang et al. 2011).

3.2 Elution chromatography experiments

Elution chromatograms of Mn00, Mn01, Mn02, and Mn03 are shown in Fig. 3.

Fig. 3
figure 3

Elution chromatograms of the alumina and the modified samples

Full size image

In these elution chromatograms, the first high and sharp peak represents the solute He that hardly adsorbs on the alumina, and its retention time is considered as the column’s hold-up time, which is always labeled as tM (Mcnair and Miller 2009). The following three peaks are identified as H2, HD, D2, the retention times of which are denoted as t R1, t R2, t R3, respectively. The distance between maxima of the i, j peaks, d ij , is calculated as

$$d_{ij} = t_{\text{Rj}} - t_{\text{Ri}} \, ({\text{j}} > {\text{i, i}} = 1,2,{\text{ j}} = 2,3)$$
(1)

The resolution between peak i and peak j, R s(ij), is defined as

$$R_{\text{s}} (ij) = \frac{{2d_{ij} }}{{(w_{\text{bi}} + w_{\text{bj}} )}}$$
(2)

w bi, w bj is the i, j peak width at baseline, respectively. As shown in Fig. 3, the resolution between H2 and any other compound for Mn00 can hardly be determined due to the poor H2 peak shape. All peaks of the modified samples are baseline separated except separation between H2 and HD peak in Mn03, and the peak width at half height is used for resolution calculation. R s (ij) of the modified samples are shown in Fig. 4. R s(12), R s(13), R s(23) show similar trends. R s is relatively low and hardly determined when Mn content is zero, it increases after the deposition, and then keeps flat, at last declines with increasing of Mn content.

Fig. 4
figure 4

Chromatographic resolutions of the alumina and the modified samples

Full size image

R s change can be affected by three parameters, namely, retention factor, separation factor, number of theoretical plates (Mcnair and Miller 2009). So it is worthwhile to discuss the relationship between R s and these parameters.

The retention factor, k i, is the ratio of the amount of solute in the stationary phase(solid phase) to the amount in the mobile phase (gas phase). According to the elution chromatogram, it is calculated as

$$k_{\text{i}} = \left( {\frac{{t_{\text{Ri}} }}{{t_{\text{M}} }}} \right) - 1 \, ({\text{i}} = 1,2,3)$$
(3)

The separation factor, α ij, is defined as

$$\alpha_{\text{ij}} = \frac{{k_{\text{j}} }}{{k_{\text{i}} }} \, ({\text{j}} > {\text{i}}, {\text{i}} = 1,2,{\text{ j}} = 2,3)$$
(4)

w bi is calculated as

$$w_{\text{bi}} = 4 t_{\text{Ri}} /\sqrt {N_{\text{i}} }$$
(5)

N i is the number of theoretical plates.

If adjacent peaks such as the H2 and HD, are assumed to have the same peak widths, i.e. wb1 = wb2, the Eq. 2 can be simplified as

$$R_{\text{s}} \left( {12} \right) = \frac{{d\left( {12} \right)}}{{w_{{{\text{b}}\left( 1 \right)}} }}$$
(6)

and R s(12) can be rearranged to form the equation

$$R_{\text{s}} (12) = \left( {\alpha_{12} - 1} \right)\left( {\frac{{k_{1} }}{{k_{1} + 1}}} \right)\left( {\frac{{\sqrt {N_{1} } }}{4}} \right)$$
(7)

The relationship between the parameters and the resolution can be given by Eq. 7. α 12, k 1, N 1 of the modified samples are calculated based on the equations above and shown in Table 2. R s(12) in Table 2 is calculated according to Eq. 7.

Table 2 Elution parameters of the modified alumina samples
Full size table

In Table 2, α 12 slightly decreases as MnCl2 content increases, it is estimated that Mn00 would have the largest separation factor among the samples. k 1 decreases with increasing MnCl2 content, Mn00 would have the largest retention factor. So poor R s(12) of Mn00 would result from poor N 1. According to the van Deemter Equation, N is related to three effects, i.e. eddy diffusion (the A term), longitudinal molecular diffusion (the B term), and mass transfer in the stationary phase (the C term).The MnCl2 deposition would mainly change the C term and improve solute adsorption and desorption kinetics which keep the same molecules close together and reduce the band broadening.

The nearly same resolution is observed between Mn01 and Mn02, higher k 1 but lower N 1 is observed in Mn01. The resolution of Mn03 with the MnCl2 content more than 30 wt% is lowering mainly due to decreasing k 1 and N 1. N 1 decreasing may result from k 1 which is a main factor in the C term (Mcnair and Miller 2009). As k 1 can affect α 12 and N 1, it would be a crucial factor to influence the resolution.

3.2.1 Relationship between textural properties and elution performance

According to the definition, k i is related to adsorption behavior of hydrogen isotope which is strongly affected by the pore size and the specific surface area of nanoporous adsorbents (Chu et al. 2006). The PSDs of the specific surface area SDFT of Mn00, Mn01, Mn02, Mn03, which are computed by DFT model, are shown in Fig. 5. Coexistence of micropores and mesopores shown in Fig. 5 indicates that k i contributes from both micropores and mesopores. To study individual contribution from different pore size range, SDFT is divided into three sections (microporous range, mesoporous range, macroporous range), the total specific area of each section is given in Table 3.

Fig. 5
figure 5

PSD of SSA of the alumina and the modified samples modeled by DFT

Full size image
Table 3 SSA of the alumina and the modified samples calculated by DFT
Full size table

As shown in Fig. 5, micropores with pore size near 0.7, 1.5 nm exist in the Mn00. The pores of near 0.7 nm were eliminated after the deposition, and the microporous SDFT decreases from 26.9 to 10 m2/g as shown in Table 3. At the meantime, the separation resolution is significantly improved. Adsorption of hydrogen in micropores is enhanced by the overlapped adsorption potential which does not occur in mesopores (Chu et al. 2006). This enhancement may be the cause of long retention time and poor resolution for elution of hydrogen isotopes by 5A zeolite (Whisnant et al. 2011). The enhancement is also observed in Mn00 compared to the modified samples, and it would origin from the micropores near 0.7 nm. It is interesting that the enhancement would be significantly weakened when the micropore size increases to near 1.5 nm.

As the MnCl2 content increases, k 1, α 12 decreases when the microporous SDFT changes little but the mesoporous SDFT decreases from 77.0 to 43.7 m2/g. This trend infers that increasing mesoporous SDFT would be helpful for resolution improvement. As the coexistence of micropores and mesopores is still observed in Fig. 5 after the deposition, it is hard to distinguish which size range is more effective for the elution performance. The ordered mesoporous materials (OMMs) such as MCM-41 (Rouquerol et al. 2012), would be helpful to understand the nature of hydrogen isotope elution. In OMMs, micropores can be ignored, relative large specific area is expected to improve k, α, and relative narrow PSD will aid to determine the effective range.

4 Conclusions

  1. (1)

    According to the DFT analysis, the activated alumina contains mainly mesopores with wide size range and micropores with narrow size range. After MnCl2 deposition both the pore volume and the specific surface area is lowered, micropores with 7 nm size disappeared. The microporous SDFT (pore size is mainly 1.5 nm) remains as 10 m2/g, mesoporous SDFT decreases from 120.8 to 43.7 m2/g.

  2. (2)

    In the elution chromatogram of the unmodified alumina, t R is relatively long, and R s is relatively poor. After MnCl2 deposition, the retention time of H2, HD, D2 deceases, R s (H2-HD) rises to 1.3–1.8, R s (HD-D2) rises to 2.0–3.0, and R s (H2-D2) rises to 3.0–4.5.

  3. (3)

    The column resolution is greatly affected by the PSD of SDFT. Poor resolution of the column packed with unmodified alumina may be caused by micropores that have relatively strong adsorption potential compared with mesopores.