Field of the Invention The present invention relates to a process for pickling metals in acid solutions containing high concentrations of hydrogen peroxide. Means is provided to recover spent pickling solution that has become contaminated with dissolved metals.

Background of the Invention.

During finishing of stainless steel, the steel typically is rolled and then annealed to achieve the desired structure and material properties. Because annealing is carried out in the presence of air, a film of oxide forms on the surface of the steel.

The film of oxide is fairly porous and contains small cracks. A zone that is depleted of chromium is normally formed under the oxide film.

A process called pickling is used to clean and condition the surface of the metal after annealing. The pickling of stainless steel requires three distinct processes.

The first is removal of the thermally grown scale for appearance purposes and to facilitate cold working of the steel. The second process increases the corrosion resistance of the final product by dissolving the chromium-depleted layer. During the third process, a minimum amount of bulk steel is dissolved, giving the desired brightening effect to the final product.

The first, scale removal step is often accomplished using shot blasting or electrolytic pickling in neutral salt. The remaining two pickling steps are traditionally carried out in mixed acid solutions which typically contain 90-160 g/L nitric acid and 10-40 g/L hydrofluoric acid As the steel is pickled, ferric iron in solution (Fe3+) reacts with iron to produce ferrous iron (Fe2+) according to equation (1).

2Fe3+ + Fe o 3Fe2+ (1) This Fe2+ is immediately oxidized to Fe3+ by the nitric acid present according to the reaction (2) below: NO3-+ 3Fe2+ + 4H+- 3Fe3+ + 2H20+ NO (2) The Fe generated by this reaction then reacts with the fluoride from the hydrofluoric acid forming complexes such as FeF3 and FeF2+,

A byproduct of reaction (2) is nitrogen oxide gas (NO) which is emitted from the solution into the atmosphere. Control of fuming from mixed acid pickling baths is a major environmental issue. Various techniques are employed to control these emissions including water scrubbing, caustic scrubbing, urea addition, selective catalytic reduction and hydrogen peroxide addition. Because of the difficulty in controlling nitrogen oxide emissions, a considerable effort has been invested by the industry in developing alternative'nitrate-free'pickle bath chemistries that do not generate an air pollution problem.

A further disadvantage of the mixed acid pickling process is the generation of spent pickle liquors containing residual concentrations of nitrate. Water pollution control by treatment of this nitrate is also problematic and a major disadvantage of the mixed acid pickling process.

Various pickling processes such as the Cleanox@ process from Henkel, have been developed that use sulphuric acid, hydrofluoric acid and hydrogen peroxide, totally eliminating nitric acid. l234. In this so-called'nitrate-free'pickling bath, dissolution of the metal still occurs according to reaction (1) and fluoride is used to complex the resulting ferric ions. The major difference is that hydrogen peroxide is <BR> <BR> used to re-oxidize the ferrous iron back to the ferric form (see equation (3) ) instead of nitric acid.

2Fe2+ +H202 + 2H+@ 2Fe + H20 (3) Unfortunately, hydrogen peroxide is very unstable under normal pickling bath conditions and can auto-decompose according to equation (4): 2H202 + °2 + 2H20 (4) Because of the high rate of auto-decomposition under normal pickling process conditions, it is not feasible to maintain a excess of peroxide in the solution. Nitrate- free processes such as the Henkel Cleanox process utilize peroxide to re-oxidize the ferrous iron to ferric, but do not operate with excess peroxide in the bath. Although the details are beyond the scope of this invention, there are conditions where the pickling process would be enhanced if an appreciable residual of peroxide were present in the pickling solution.

Summary of the Invention The present invention is concerned with nitrate-free stainless steel pickling baths that contain a significant concentration of hydrogen peroxide.

According to the present invention there is provided a process for pickling metal surfaces using solutions containing hydrogen peroxide and acid wherein the pickling solution, which has accumulated dissolved metal contamination, is cooled to a temperature less than the operating temperature of the pickling process and then fed to a nanofiltration process. The nanofiltration process is used to concentrate the metals in the rejected pickling solution. The permeate solution from the nanofiltration process, which has a reduced metal concentration, is recycled back to the pickling process for reuse.

In a preferred embodiment of the invention, the temperature of the solution being fed to the nanofiltration process is cooled to a temperature of less than 10°C.

By cooling the pickling solution to a temperature of less than 10°C, the corrosiveness of the solution is reduced so that hydraulic components of the apparatus can be fabricated in an austenitic stainless steel alloy such as 316L stainless steel.

In order to improve the recovery efficiency of acid, the reject solution from the nanofiltration process can optionally be further processed by a sorption unit such as an acid retardation unit or a diffusion dialysis unit, capable of sorbing the acid values contained in the nanofiltration reject solution. The acid that is sorbed is then recovered from the sorption unit by desorption with water. To avoid oxidation of the adsorption media, the residual hydrogen peroxide in the reject must be reduced.

Nanofiltration membranes and most common construction materials are not chemically resistant to peroxide containing pickling solutions. Moreover, if dissolved metals are concentrated in the pickling solution by a process such as nanofiltration, the peroxide decomposes at an unacceptably high rate due to the catalytic effect of the metal ions.

By reducing the temperature of the spent pickling solution below normal pickling process temperatures, it has been found that the chemical attack on the membranes can be substantially reduced, while at the same time, the stability of the peroxide is maintained, even at high metal concentrations. An additional benefit is that at the low temperatures in which this invention is operated, the corrosivity of the solution is substantially reduced so that low cost materials of construction such as 316L stainless steel can be utilized for hydraulic components such as pumps, vessels and piping.

Brief Description of the Drawings In the accompanying drawings: Figs. 1 and 2 are graphs illustrating hydrogen peroxide decomposition over time. Fig. 1 shows the results of an experiment that was conducted to show the effect of iron concentration on peroxide decomposition at a normal pickling bath temperature and Fig. 2 is a similar graph showing the results of a laboratory test in accordance with the process according to the invention; and, Fig. 3 is a schematic illustration of an apparatus according to a preferred embodiment of the invention.

Description of Preferred Embodiment Catalytic auto-decomposition of peroxide due to the presence of dissolved metals such as iron is well-known. Ferric iron, is particularly effective. According to Schumb5, at high levels of excess peroxide in relation to ferric iron, the rate of decomposition is a function of the ferric iron concentration of the solution according to the expression: - dCHzO/d = k (Fe) (H202)/ OH )

At iron concentrations of 20-30 g/L such as normally employed, the decomposition rate is very high, however if the iron concentration is reduced to less than 10 g/L, the decomposition rate slows considerably. An experiment was conducted to show the effect of iron concentration on peroxide decomposition at 45°C, which is a normal pickling bath temperature. The initial solution had a composition of: Ferric iron = 5,15 and 30g/L Sulfuric acid = 40 g/L Hydrofluoric acid = 20 g/L Hydrogen peroxide = 30 g/L The results of this experiment are shown in Figure 1. It can be seen that the peroxide is extremely unstable at an iron concentration of 30 g/l, the concentration being reduced to about 10% of its initial level in about one hour. At an iron concentration of 15 g/L, the rate of decomposition was significantly reduced. A 10% decrease occurred in a about 4 hours. At an iron concentration of 5 g/L no significant decomposition of the peroxide occurred.

In order to maintain the pickling bath at a constant iron concentration it is necessary to continually bleed-off pickle liquor and replace it with fresh solution.

Obviously the quantity of spent pickle liquor generated varies inversely with the allowable maximum iron concentration. The advantage of improved peroxide stability at lower metal concentration is obviated by the increased cost of spent pickle liquor disposal and replacement chemicals, including peroxide.

Equation (5) also implies that the stability of peroxide is improved at higher acid concentration. The potential advantage of improved peroxide stability by use of higher acidity solutions is also obviated by the increased cost of acid neutralization of the spent pickle liquor.

It is apparent that the economic feasibility of operating a pickling bath in this way is predicated on some form of spent pickle liquor recovery system that would recover the residual free acid and peroxide values and recycle them back to the pickling process.

The industry has extensively utilized acid retardation systems such as the Eco- Tec APU system for many years to recover mixed acid pickle liquors. Indeed, the same system has been used to recover Cleanox nitrate-free pickle liquors with good success. The size and cost of these types of recovery systems is flow dependent. A recovery system designed to maintain a pickle bath at 10 g/L of iron would have to be about three times as large as one designed to hold the bath at 30 g/L. Consequently, the economics for operating a pickle bath in the peroxide-stable regime (i. e. Fe <10 g/L) are very unfavourable.

Recently, nanofiltration (NF) membranes have become commercially available which are capable of separating dissolved metal contamination from strong mineral acids. In the NF process, spent pickle liquor is pumped past the membrane surface at high pressure. The membrane allows free acid to permeate through, while metal salts are largely rejected. The collected permeate contains free acid at about the

original feed concentration and a metal concentration at about 10-30% of the initial level. The permeate can be recycled back to the pickle bath. The reject also contains free acid at about the original bath concentration, however the metal concentration can be increased several fold above the original level. The reject is typically discarded as waste. Since the acid to metal concentration ratio in the reject is much less than the feed solution, the NF system can be a reasonably effective recovery system under many circumstances.

The NF process has a number of advantages over the acid retardation process for this application. First of all, because of its ability to concentrate the metals, the flow rate of waste solution can be significantly less that the feed. With acid retardation, the flow rate of the waste is typically the same or greater than the feed flow. In addition, because of lower osmotic pressures and corresponding higher fluxes at low metal concentration, NF has the ability to treat solutions containing relatively low metal concentrations. Acid retardation systems on the other hand are sized based upon feed flow, independent of metal concentration. Generally speaking, acid retardation systems are more attractive at high pickle bath metal concentrations and NF systems are more attractive at low metal concentrations.

Temperature has a well-known effect on pressure driven membrane processes such as reverse osmosis and nanofiltration. Generally speaking, higher temperatures provide higher flux rates, thereby reducing the amount of membrane area and the capital cost of a system designed to process a given amount of liquid. The only advantage to processing at lower temperatures is a marginal improvement in rejection efficiency. In cases where fluid temperatures are less than 10°C, it is standard practice to heat the solutions up to 20-25°C prior to treatment, to improve flux rates.

Cooling solution temperatures to well below normal ambient temperatures is not usual practice, since there are no significant advantages to doing so.

As discussed above, in order to stabilize the peroxide, it is necessary to operate a peroxide pickling bath at low metal concentrations. This would tend to favour use of an NF system instead of an acid retardation system. Unfortunately, because the NF system concentrates the metal, this would promote peroxide decomposition during treatment and obviate the advantage of the NF process.

It is well known that peroxide decomposes much more quickly at elevated temperatures. For example, according to Schumb6, the rate of decomposition increases by a factor 2.2 for a 10°C rise in the range of 50-60°C. Little data is available on peroxide decomposition at temperatures below ambient levels (i. e.

<20°C), however. Tests were conducted similar to those described above at ambient temperature (i. e. 20-25°C). The results are summarized in Figure 2. It can be seen that the peroxide stability is appreciably enhanced by reducing the temperature. By reducing the solution temperature from a typical pickling bath temperature of 45°C to ambient, the adverse effect of elevated metals on peroxide stability can be effectively counteracted. Even at an iron concentration of 30 g/L, the peroxide concentration reduction was less than 5% after four hours. According to the present invention, the spent pickle liquor is reduced in temperature, below normal pickling bath temperatures before being processed by the nanofiltration process. This avoids decomposition of the peroxide as the metal concentration is increased in the NF process.

While the basic NF process has been around for many years the availability of suitable acid resistant membranes is very recent. Examples of acid resistant NF membranes include MPT-34 from Koch Membranes Systems, DS-5 from Osmonics and N30F from Nadir Systems. While these membranes have good tolerance to relatively high concentrations of acids such as sulfuric acid, their resistance to strong oxidizing agents is limited. In fact, neither the MPT-34 or DS-5 membranes are recommended for hydrogen peroxide service by their manufacturers. The Nadir N30F membrane, which is a polyethersulfone membrane, was reported by the manufacturer to be resistant up to 35,000 mg/L of free chlorine, a strongly oxidizing environment.

No data was available on resistance to peroxide, however.

Conventional NF membranes used in water treatment applications are normally not suitable for use above about 35°C. Acid resistant NF membranes on the other hand, are typically resistant to much high temperatures. For example, Nadir N30F is good up to 95°C. This is well above normal pickling process temperatures, which are typically 40-60°C. Therefore it is not usual to cool pickle liquors before NF treatment.

In order to assess the resistance of the N30F membrane to peroxide, an experiment was conducted. Samples of the N30F membrane were immersed in a simulated pickling bath of the following composition: Ferric iron = 28 g/L Sulfuric acid = 75 g/L Hydrofluoric acid = 28 g/L Hydrogen peroxide = 24 g/L The experiment was conducted at ambient temperatures (20-25°C) and at 5°C.

The membrane samples were withdrawn and tested for flux and rejection efficiency.

Tests were conducted with a 1% Na2SO4 solution at ambient temperatures. According to the manufacturer's catalog, new membrane has a flux of 40-70 1/h/m2 and a rejection efficiency of 85-95% at 580 psi. After 36 days immersion at room temperature, the flux decreased to 23 1/h/m2 and the rejection decreased to 28%. This is an indication of severe membrane degradation. Tests were also conducted at 6°C.

At this temperature the membrane was found to have a flux of 53 1/h/m and a rejection efficiency of 89.5% at the same pressure, after the same time period. This is an indication of only minor amounts of degradation, if any.

Because of the high operating pressures involved (400-800 psi) in NF, the hydraulic equipment including pump, pressure vessels and piping of an NF system represents a significant portion of the equipment cost. Normally, the preferred material of construction is an austenitic stainless steel such as alloy 316. Since the fluid being treated here is actually used to pickle 316 stainless steel, some other more corrosion resistant alloy would normally be considered. Hydraulic components fabricated in exotic alloys such as alloy 20 are much more expensive than 316 stainless steel, however. It has been found that if the temperature of the feed is reduced, the corrosion rate of 316 stainless steel can be reduced to an acceptable level.

An experiment was conducted to determine the corrosion rate of 316L stainless steel in peroxide pickle liquors. Samples of the metal were immersed in solution of. the following composition: Sulfuric acid = 75 g/L Hydrofluoric acid = 30 g/L Hydrogen peroxide = 30 g/L The dissolved metal concentration in the solution was then analyzed after a period of time. From this information, the corrosion rate can be calculated. The experiment was conducted at ambient temperatures (20-25°C) and 6°C. The corrosion rate at ambient temperature was found to be 0.052 inches per year (52 mils per year). The corrosion rate at 6°C was found to be only 12 mils per year. Normally an allowable rate of corrosion is considered to be less than 20 mils per year. This confirms that 316L stainless steel is not a suitable material at temperatures at or above ambient, however it indicates that if the temperature is reduced significantly below ambient temperatures, the corrosion rate is acceptable and the material can be used.

Unexpectedly, it has been found that nanofiltration is an excellent way to recover a peroxide containing acid pickling bath. Conventional wisdom would suggest that: 1. The increased metal concentration resulting from operation of the NF system would de-stabilize the peroxide and cause rapid peroxide degradation.

2. The membranes are not resistant to peroxide under normal pickling conditions.

3. It is necessary to use exotic corrosion resistant alloys for components to treat pickle liquor under normal conditions. This would make the system extremely expensive to fabricate.

It has been found that by reducing the temperature of the spent pickle liquor, these difficulties can be avoided.

As discussed above, normally in the NF process, water and free acid pass readily through the membrane. In other words, their rejection is very low. Metal salts typically have very high rejection efficiency. No data on the rejection of hydrogen peroxide was found in the technical literature.

An experiment was conducted to determine the rejection of the various components. Simulated pickle liquor containing sulfuric acid, hydrofluoric acid, iron and hydrogen peroxide was recirculated through a test cell fitted with a small sample coupon. of N30F membrane at a pressure of 600 psi. The results are summarized in Table 1. The feed solution was formulated to simulate the pickle liquor after it had been concentrated by the process to some degree.

Table 1 Component Feed Permeate % rejection H202 23.6 g/L 18.7 g/L 21% Total free acid 1. 58 N 1. 76 N-11% Fe 28 g/L 7.8 g/L 72% Total F 27. 8 12. 3 56% It can be seen that indeed, the rejection of the acid is very low. In fact, the acid concentration has been concentrated somewhat in the permeate an indication that the acid is actually passing through the membrane faster than the water. As expected, the iron has been rejected reasonably well (72% rejection).

The concentration of peroxide in the reject was approximately 21% lower than that of the feed solution. This represents an apparent 21% rejection. A more likely explanation was that since the test was conducted at ambient temperatures, the peroxide partially decomposed due to the presence of a relatively high iron concentration in the feed solution.

A pilot plant was assembled containing a nanofiltration membrane element manufactured by Nadir with a surface area of 4.18 m2. Acid pickling liquor containing hydrogen peroxide, sulfuric acid, hydrofluoric acid and iron was circulated through a chiller and the membrane module using a 316 stainless steel multi-stage centrifugal pump. The temperature of the solution was held at approximately 2-4°C and a back pressure of 518 psi was maintained on the membrane module. The average permeate flux was 10.6 L/m2/h. The results after approximately 30 minutes of operation are summarized in Table 2. No reduction in the hydrogen peroxide concentration in the feed solution was observed over the duration of the experiment.

In addition, as can be seen from the data, there is no rejection of peroxide by the membrane.

Table 2 Feed Reject Permeate Rejection [Fe] (g/L) 25.96 25.69 9.33 63.7% [H] f (N) 1.64 1.58 1.63-3. 2% [H202] (g/L) 21.9 20.7 21.9-5. 8% The pilot plant NF tests confirmed the laboratory peroxide stability experiment of Figure 2. It was noted however, that when the metal concentration was increased much beyond 30 g/L, for example to approximately 50 g/L, the peroxide rapidly degraded. As will become apparent, this is actually an advantage.

U. S. patent 5,547, 5797 by the present inventor shows a way to integrate an acid sorption process such as acid retardation or diffusion dialysis with nanofiltration in order to increase the recovery efficiency of free acid. Diffusion dialysis (DD) utilizes an ion exchange resin very similar to that employed in the acid retardation process. In the DD process the ion exchanger is in the form of a membrane, while in the acid retardation process, the ion exchanger is in particulate form. The processes

are analogous to each other in that acid is first sorbed into the ion exchanger and then extracted or eluted from the ion exchanger with water. According to a preferred embodiment of the present invention, the reject from the nanofiltration process is treated by a sorption unit (ASU) such as an acid retardation unit like the Eco-Tec APU or a diffusion dialysis unit, in a similar way to that taught by U. S. patent 5,547, 579, to improve the recovery efficiency of the acid.

In the preferred embodiment of this invention, the NF reject stream feeding the ASU contains sulfuric acid, at about pickling bath concentration or slightly lower, and metal salt at an appreciably increased concentration. The acid sorption unit will sorb the acid values. The resulting solution leaving the ASU, normally called the 'byproduct', will contain high concentrations of metal but very low concentrations of acid. The byproduct normally is considered waste. Water desorption of the ASU media will recover a solution called the'product', which contains high concentrations of acid. The ASU product could be recycled back to the pickle bath. Since the ASU product will contain small, but significant concentrations of metal, it could alternatively be fed back into the NF system. By this means, the overall recovery of acid and peroxide can be significantly increased compared to the NF system alone.

Sorption media such as the ion exchangers employed in acid retardation and diffusion dialysis units are susceptible to oxidation by hydrogen peroxide. An experiment was conducted to determine the oxidation resistance of the ion exchange resin employed in an Eco-Tec APU to the reject solution from the nanofiltration process. The resin was immersed in a beaker containing a simulated NF reject solution with the following initial composition: [Fe] = 34 g/L [HF] = 18. 5 g/L [H2S04] = 49 g/L [H202] = 30 g/L The temperature of the solution was held at 3-5°C. It was found that the peroxide concentration rapidly depleted, dropping from 30 g/L to 12.9 g/L in 72.5 hours. Apparently, the presence of the ion exchange resin in the solution caused the destruction of the peroxide, since this rate of depletion was much more rapid than the previous experiment whose results are summarized in Figure 2.

This experiment was conducted over a period of 498.5 hours. The peroxide concentration was analyzed from time to time and replenished to approximately 30 g/L. After 311 hours, the moisture content of the resin was found to increase from an initial value of 45.5% to79.1%. This is an indication of severe oxidation. Normally ion exchange resins of this type are considered spent when the moisture content exceeds about 60%. After 498 hours the moisture content increased to 91.1%. It is evident that in order to incorporate an ASU into the flowsheet as taught by the'579 patent, it is necessary to ensure that peroxide has been removed from the NF reject stream which forms the feed to the ASU.

Figure 3 is a schematic illustration of an apparatus according to a preferred embodiment of this invention. Stainless steel strip 1 is passed through a pickle bath comprising a tank 2 holding a peroxide containing pickling solution 3. As a result of chemical attack of the pickling solution on the metal, the concentration of dissolved

metal in the pickling solution increases. Pickling solution is withdrawn from the pickling tank at 4 and its temperature is reduced by a chilling device 5. The cooled pickling solution is then passed via line 6 to a nanofiltration system 7. The permeate from the nanofilter containing a reduced metal concentration is recycled back to the pickling tank via line 8. As the metal is concentrated in the NF unit from an initial level of <10 g/L (say 5 g/L) up to about 30 g/L, the peroxide remains stable and the permeate will contain peroxide at about the same level as present in the spent pickle liquor. The peroxide will begin to degrade however as the metal concentration is increased from say 30 g/L to 50 g/L. The permeate will not contain much peroxide at this point, however since by this time most of the peroxide has been recovered. The final NF reject has a high metal concentration say 30-50 g/L, but will contain little peroxide. Any residual peroxide can be removed by dosing with a chemical reducing agent such as sodium sulfite. This reject solution is passed via line 9 to an acid sorption unit 10. Alternatively the reject solution from the NF can be held for a time in a tank and allowed to warm up a little, ensuring that any remaining peroxide will auto-decompose, so that the solution fed to the ASU will contain no peroxide. As a result, no chemicals will be required to destroy peroxide.

The acid in the solution fed to the ASU is taken up by the absorption media in the acid sorption unit and resulting byproduct solution 13 is collected as waste. Water 11 desorbs acid from the acid sorption unit media and a product solution is collected and passes via line 12 back to the pickling tank. Alternatively, this product solution, which still contains some metal can be combined with the spent pickling solution and fed to the nanofiltration system again.

A significant portion of the operating cost of the system is associated with the energy to operate the chilling device. It will be apparent that the amount of energy will be reduced by using a heat exchanger in such a way so that the permeate from nanofiltration pre-cools the spent pickle liquor before passing to the chiller.

References 1. Zavattoni, M. , U. S. patent 5,785, 765, July 28,1998.<BR> <P>2. Henriet, D. , U. S. patent 5,690, 748, November 25,1997.

3. Japan patent 243289/85.

4. European patent DE 2,827, 697.

5. Schumb, W. C. , et al, "Hydrogen Peroxide", p. 492, ACS Monograph Series, Reinhold Publishing Company, New York.

6. Schumb, W. C. , et al, "Hydrogen Peroxide", p. 525, ACS Monograph Series, Reinhold Publishing Company, New York.

7. Brown, Craig J. , U. S. patent 5,547, 579, Aug. 20,1996.