Separation of CO-CO,-N, gas mixture for
high-purity CO by pressure swing
adsorption
Li-qun Zhu*, Jin-ling Tu and Ya-jun Shi
Chemical Engineering Research Centre, East China University of Chemical Technology,
Shanghai 200237, China
A pressure swing adsorption (PSA) process for separating CO from a CO-C02-N2 mixture is proposed.
The adsorbent used in this process is active carbon supported copper, which has been developed by
this laboratory. By cycling the pressure of a bed of this adsorbent between ambient pressure and
20-30 Torr at room temperature, high purity CO can be obtained from the CO-C02-N2 gas mixture with
a high recovery. The CO product purity depends crucially on the step of CO cocurrent purge after
adsorption in the cycle and the regeneration of sorbent.
Keywords: CO separation; PSA; active carbon; adsorption
Introduction
Efficient methods for separating carbon monoxide (CO)
from gas mixtures are important in industry, since almost
all CO sources are gas mixtures with nitrogen, hydrogen,
methane and carbon dioxide’. Many liquid adsorbents for
CO have been proposedza. However, the energy require-
ments of absorption are high and the solvents are easily
deactivated by trace amounts of polar gases. As the new
solid adsorbents for CO were developed successfull$.6,
separation of CO from gas mixtures by pressure swing
adsorption (PSA) processes became possible.
Gas separation by PSA technology has become an
important unit operation in the chemical industry in
recent years. The major applications of the technology
involve air drying, air separation, octane improvement
and H, purification 7-9 CO recovery from blast furnace .
gases by PSA technology was first developed jointly by the
Kawasaki Steel Corporation and Osaka Oxygen Ind. Ltd.
The process consists of two stages of separation. The first
is PSA for carbon dioxide removal by active carbon and
the second is for CO recovery where sodium-type
mordenite is used”. Mitsubishi Kakoki Kaishi Ltd. also
developed a PSA process by using a CO-selective
adsorbent developed by Hirai er al.’ for producing CO”. In
this paper, a PSA cycle for separating CO from a CO-
C02-Nz mixture by using active carbon supported copper,
the preparation of which6 was developed in this labora-
tory, is proposed. This cycle does not require a pretreat-
ment bed and can give a high-purity CO product from
CO-CO*-N, with a high recovery.
*Present address: No. 35 Hua Yan Bei Road, West Building, Beijing,
100083, China
Process description
The CO-selective adsorbent used was prepared by
supporting cupric chloride on active carbon and then
heated in nitrogen at 300°C. This preparation method was
developed in this laboratory. This active carbon supported
copper obtained by this method has a high CO adsorption
capacity and a satisfactory selectivity. It can separate CO
from CO-C02-N, gas mixture efficiently.
The proposed PSA cycle involves four steps: (1)
pressurization, (2) adsorption, (3) CO cocurrent purge and
(4) countercurrent evacuation. A detailed description
of each step follows.
Adsorption
The feed gas flows through the adsorption bed at feed
pressure. The bed, which contains active carbon supported
copper, has been previously pressurized to feed pressure
with a nitrogen-rich gas or feed gas. The feed pressure in
this process is ambient pressure. The effluent from this
step constitutes a N,-rich gas, a part of which is taken as
the N, product and the remainder used for the bed
pressurization step described later. Adsorption is con-
tinued until the CO* breaks through completely.
CO cocurrent purge
To increase the CO product purity, which is a more select-
ively adsorbed component, a CO cocurrent purge step
after adsorption has been introduced. This step not only
enriches CO in the void spaces of the bed, but also dis-
places the CO? and Nz which have been adsorbed on
0950-4214/91/030173-04
0 1991 Butterworth-Heinemann Ltd Gas Separation 8 Purification 1991 Vol 5 September 173
PSA process for separating CO from CO-CO,-N, gas mixture: Li-qun Zhu et al.
sorbent by CO. The effluents in this step are recycled to
the feed mixture.
Countercurrent evacuation
When the CO cocurrent purge step finished, the bed was
evacuated countercurrently by a mechanical vacuum
pump. A vacuum desorption step must be used because
CO is a strong adsorbate. The evacuation step serves to
obtain CO product and to clean the bed. So effecting
separation of the CO-C02-N2 mixture was possible. Some
of the gas obtained from this step is collected as CO
product and the rest used as the purge gas in step 3.
Pressurization
At the end of the evacuation step, the pressurization step
begins. The bed is pressurized to the feed pressure level by
introducing some of the N,-rich gas produced during step
2 countercurrently or feed gas cocurrently. The bed is now
ready to undergo a new cycle starting from step 2.
Feed
gas
PU
Pressurization
u
Figure 2 Schematic diagram of apparatus for PSA for CO-C02-N,
separation: I, adsorption bed; 2, pressure gauge; 3, manometer;
4. vacuum pump; 5. flowmeter; 6, gas volumometer; SV, solenoid
valve; TC, thermocouple: SP, sampling port
One mode that contains three adsorbent beds has been
proposed for operating this four-step process. Figure I
shows a schematic flow diagram for this arrangement.
This configuration allows continuous feed gas flow and
product gas withdrawal. The performance of a PSA
process is determined by three separation results: product
purity, product recovery and throughput, all at cyclic
steady-state. The product purity is defined as the volume-
averaged concentration over the entire product. The
throughput is defined as the total feed volume (l(STP))
treated per unit time per unit weight of sorbent (l(STP) h-’
kg-‘). The product recovery is defined as:
Percentage CO = (CO from step 4) - (CO used in step 3)
recovery (CO in feed) - (CO effluent from step 3)
Experimental
In order to simplify the experiment in the laboratory, a
single-bed apparatus for simulating the multibed process
was designed. A schematic diagram of the apparatus is
shown in Figure 2. The adsorption column was packed
with 20-40 mesh active carbon supported copper, which
was made by the method developed in this laboratory.
Table 1 shows the sorbent and bed characteristics. Five
solenoid valves located at the feed, purge, vent, evacuation
Table 1 Characteristics of adsorption column
Column height, L 27.7 cm
Column inside radius, r 2.632 cm
Particle size, dp 0.0675 cm
Bulk density, pe 0.5858 g cmm3
Interparticle void fraction S 0.494
Total void fraction 6, 0.693
Heat capacity of particle C, 0.2 cal g-’ K-’
and pressurization line were used alternately to direct the
flow into and out of the column. The pressure and tem-
perature history for each experiment were recorded. Gas
samples were taken from the sampling ports by syringes
and analysed by gas chromatography (GC). A mechanical
vacuum pump was used for the evacuation step. The
premixed feed mixtures contained 66.4% CO, 25.7% Nz
and 7.9% CO,.
N,, CO?, CO
A B C
A typical PSA run proceeded as follows: the bed was
cleaned before each run by heating and degassing. Step 1
was initiated by opening the solenoid valve (SV) connected
to the pressurization gas. When the bed pressure achieved
ordinary pressure, the SV connected to the feed line and
vent line were opened simultaneously and step 2 started.
Step 3, CO cocurrent purge, was effected by simultaneously
closing the feed valve and opening the purge valve. The
purge valve and vent valve were closed and the counter-
current valve was opened simultaneously to achieve step
4. The flow rate in cycle was adjusted by a line needle
valve. Acyclic steady-state was generally reached after live
cycles from start-up with a clean bed. After the steady-
state was achieved, a complete set of flow rates and
product samples were taken. The feed pressure was
ordinary pressure and the end pressure of step 4 was
20-30 Torr.
Results and discussion
Typical results for separation of CO-CO,-N,
Figure 1 The proposed PSA process for separation of CO-CO,-N1
Changes of typical outlet gas compositions with time are
shown in Figure 3. Only N, came out for the first few
minutes, and then CO, came out abruptly. This sudden
increase of CO, content, which was several times more
than that of the inlet gas, was due to the result of substitute
174 Gas Separation 8 Purification 1991 Vol 5 September
4
3
0”
t
2
I
0
PSA process for separating CO from CO-C02-N, gas mixture: Li-qun Zhu et al.
2 4 6 0 IO 12 14
Time (min)
Figure 3 Changes of typical outlet gas compositions with time
adsorption of CO onto sites first occupied by COz. This
CO* breakthrough was followed by a gradual CO break-
through within a few minutes. This figure indicated that
CO can be separated efficiently from the CO-COz-N,
mixture by using this active carbon supported copper.
For the PSA cycle, the detailed experimental results
and the effluent concentration of a typical run are
presented in Table 2 and Figure 4. These are the cyclic
steady-state results, which are accomplished approxi-
mately after five cycles from start-up. The results showed
that a high-purity CO product with a high recovery could
be achieved in PSA by using this sorbent. The variation of
effluent flow rate and concentration during step 2 and step
3 indicated the displacement of CO, and N, by CO. It was
found that step 3 is necessary for a high-purity CO
product.
Effects of ambient temperature
Generally, low temperatures are suitable for adsorption
and high temperatures for desorption. In the PSA cycle,
Table 2 Steady-state results of typical run of CO-NZ-CO2 separa-
tion by PSA
Step Time (s) yco (W Ycoz (%) Flow rate (ml) (STP)
60 2.94
2 120 3.70
180 5.32
00.: 0:04
137.8
167.4
177.2
240 9.04 35.54 285.5
300 57.47 9.74 137.8
3 360 55.02 9.59 68.9
420 75.67 20.15 118.1
480 80.68 14.57 147.7
490 97.84 0.0 196.9
510 98.01 0.0 1004.2
4 540 98.71 0.01 354.4
600 98.98 0.86 256.6
720 99.17 0.83 206.7
2 avg. CO cont. 5.91 6.62
Total amount 767.9
3 avg. CO cont. 68.92 13.83
Total amount 472.6
4 avg. CO cont. 98.58 0.12
Total amount 2010.2
Percentage CO recovery 95.46
I l * l
< I
Adsorption ’ Purge l
4
Evacuation
00.6 d I-
c
U * l
\
0 0.4
0.2 t
b h A 4
. l g , IAA A
I 2 3 4 5 6 7 6 9 IO II 12
Time (min)
Figure 4 Effluent concentration of gases in steady-state PSA:
0, co; a, co2
Table 3 Effects of ambient temperature
1 Pressure (MPa) 0.13
i Temperature (K) 303.2
CO feed concentration (%) 66.4
CO2 feed concentration (%) 7.9
Cycle time (min) 12
P/F, 0.791
Throughput (I(STP) (h-’ kg-‘))\ 107.59
Product gas CO concentra-
tion (%) 97.26
CO1 concentration (%) 1.02
CO recovery (%) 95.53
0.13 0.13
313.2 323.2
66.4 66.4
7.9 7.9
12 12
0.807 0.796
111.46 109.83
98.58 98.82
0.13 0.00
95.46 87.95
the net adsorption capacity of adsorbent can be affected
strongly by temperature. It has been found that the effect
of temperature on isotherms of CO, CO, and N2 on active
carbon supported copper is contrary12 to the norm. So,
increasing temperature appropriately, the selectivity of
sorbent can be improved. The results in Table 3 show that
CO product purity can obviously be improved by raising
ambient temperature, while the net adsorption capacity of
sorbent is kept about constant. However, if the ambient
temperature is raised sufficiently, the separation results
become poor.
Effects of cycle time
Table 4 shows the effects of cycle time when the amount of
feed gas treated remains constant. For this condition, a
shorter cycle time corresponds to a higher throughput and
faster flow rate, which can improve the separation. How-
ever, for CO-C02-N2 mixture, the separation results
strongly depend on the regeneration of sorbent. A shorter
cycle time is not suitable for CO-C02-N, separation
Table 4 Effects of cycle time
Pressure (MPa) 0.13 0.13
Temperature (K) 313.2 313.2
CO feed concentration (%) 66.4 66.4
CO2 feed concentration (%) 7.9 7.9
Cycle time (min) 9 12
P/F, 0.785 0.758
Throughput (I(STP) (h-’ kg-‘)) 143.95 111.46
Product CO concentration gas (%) 98.58 98.58
CO2 concentration (%) 0.23 0.13
CO (%) recovery 85.81 95.46
Gas Separation Et Purification 1991 Vol 5 September 175
PSA process for separating CO from CO-C02-N2 gas mixture: ii-gun Zhu et al.
Table 5 Effects of bed pressurization by N, against pressurization
by feed
Bed pressurization By feed 8~ Nz
Pressure (MPa) 0.13 0.13
Temperature (K) 313.2 313.2
CO feed concentration (%) 66.4 66.4
CO* feed concentration (%) 7.9 7.9
Cycle time (min) 12 12
P/F,, 0.959 0.907
Throughput (I(STP) (h-’ kg-‘)) 122.23 1 12.73
Product CO concentration gas (%) 99.38 98.98
COz concentration (%) 0.12 0.00
CO recovery (%) 58.81 96.96
because it can make the regeneration of sorbent become
poor, which is controlled by the evacuation capacity of the
vacuum pump.
Effects of bed pressurization by N2 against
pressurization by feed
A comparison of pressurization by N2 and by feed is given
in Table 5, where the other conditions have been kept the
same. The data shown indicates that bed pressurization
by N2 is more suitable for CO separation from CO-CO*-N,
mixture. The main function of using N, in pressurization
is to sharpen the concentration wavefront in the bed. If
pressurization by feed gas was used, the concentration
wavefront was diffuse after the pressurization step. It
caused CO breakthrough earlier and decreased CO
recovery.
Conclusion
Separation and recovery of CO from a CO-CO*-N, gas
mixture can be achieved by PSA with active carbon
supported copper as sorbent. This adsorbent was made by
the method developed in our laboratory. The proposed
four-step PSA cycle can separate a CO-C02-N, mixture
into high-purity CO with high recovery. The CO product
purity strongly depends on the regeneration of sorbent
and on the CO cocurrent purge step after adsorption in the
cycle.
References
1
2
3
4
5
6
7
8
9
10
11
12
Kohl, A.L. and Riesenfeld, F.C. GasPun~carion 3rd Edn. Gulf
Publishing Co., Houston, USA (1979)
Deringer, H. Chimia (1947) 1 125
Haase, D.J. Chem Eng News (1975) (August 4) 52
Hirai, H., Wada, K. and Komiyama, M. Bull Chem Sot Jpn
(1986) 59 109-117
Hirai, H., Wada, K. and Komiyama, M. Bull Chem Sot Jpn
(1986) 59 2217-2225
Zhu, L.Q., Tu, J.L. and Shi, Y.J. Ranliao Huaxue Xuebao
(China) (1989) 17 284-290
Cassidv. R.T. and Holmes. E.S. Paner oresented at the Ann
AlChE-Meeting. Washington DC, USA (1 November 1983)
Keller, C.E. In Industrial Gas Separation (Ed T.E. Whyte ei al)
American Chemical Society:ACSSympSerNo223(1983) 145
Sircar, S. In Adsorption: Science and Technology (Ed A.E.
Rodrigues) Kluwer Academic Publishers (1989) 285-321
Sakuraya, T. et al Kawasaki Seiretsu Giho (in Japanese) (1985)
17 135-139
Toguchi, K. er al World Congr IZZChem Eng 8f-205(R) (1986) 829
Zhu, L.Q. PhD thesis East China University of Chemical Tech-
nology, Shanghai, China (1990)
176 Gas Separation & Purification 1991 Vol 5 September