Methods of Conserving Archaeological Material from Underwater Sites
by Donny L. Hamilton

Iron Conservation: Part II - Experimental Variable and Finals Steps


Types of Electrolytic Setups
Sodium Hydroxide
Sodium Carbonate
Carbonate Sequestering Agents
Water in Electrolytes
Effects of Electrolytes on Metal Vats
Current Density
Electrode Potentials
Outdoor Electrolysis

General Observations on Cleaning Iron Objects Electrolytically

Exposure to Anode Surface
Duration of Electrolysis
Reducing Electrolysis Time
Electrolysis Summary
Alkaline Sulfite Treatment
Chemical Cleaning

Tannic Acid

Oxidizing Atmosphere
Reducing Hydrogen Atmosphere
Hydrogen Plasma Reduction
Water Diffusion
Final Conservation Steps
Rinse After Treatment
Qualitative Test for Chlorides
Sealant and Consolidation
Storage and Periodic Inspection


The manner in which artifacts are set up for electrolysis is dependent upon the following factors (Hamilton 1973, 1976):

  1. the size and condition of the specimens
  2. the amount of artifacts to be processed
  3. the number of available regulated direct current power supplies
  4. the current capacities of the power supply units
  5. the number, size and nature of the vats

Fig 10B

The 'ideal' electrolytic setup (Figure 10B.1A) consists of a single artifact in one vat, surrounded by a close, form-fitted anode that is equidistant from all surfaces of the artifact and is connected to a single regulated direct current (DC) power supply. With this setup, a conservator is able to precisely regulate the flow of the current to the artifact and maintain a predetermined electrode potential conducive to metal reduction on the surface of the specimen. This setup is used for artifacts that are especially significant and need to be conserved as carefully as possible.

The Type 2 electrolytic setup (Figure 10B.1B has several artifacts in one vat, but each artifact is surrounded by its own close, form-fitted anode, and each is connected to a separate DC power supply. When this arrangement is used, it is important to make sure that the distance between the different anodes is greater than the distances between the artifact and its anode in order to prevent any cross-over current. (This point is not graphically depicted in Figure 10B.1b.) With this setup, the current flow to each artifact can be carefully controlled, and the correct electrode potential can be maintained. Since the chlorides present in the electrolyte come from all the artifacts in the vat, it is not possible to determine exactly when a specific artifact is chloride-free. The chloride test does, however, tell the conservator when to change a chloride-contaminated electrolyte and when all of the artifacts are chloride-free.

If an artifact requires close supervision (i.e., to consolidate a metal-oxide interface or to preserve some surface or structural detail), it is advisable to hook it up in one of the two methods described above. The most critical variable for precision control is the ability to maintain an even current density on the cathode surface by positioning the anode equidistant from all parts of the artifact while maintaining a steady reduction electrode potential. Monitoring the chlorides during electrolysis is of lesser importance.

The most commonly used electrolytic setup involves connecting multiple artifacts to a single power supply. Regardless of how it is configured, Type 3 electrolytic setups (Figure 10B.1C-D) are the least desirable from the standpoint of control, but they have the advantage of processing a number of objects at one time in one vat on a single power supply. In one configuration (Figure 10B.1C), each piece is individually connected to the negative terminal of a single power supply. The artifacts share common anode sheets placed above and below the specimens. An alternative variation of this setup has a common bottom anode and individual top form-fitted anodes to assure a more even distribution of the current to each artifact.

In the most popular Type 3 configuration (Figure 10B.1D), artifacts are suspended from a brass cathode rod conductor (Plenderleith 1956:194-196; Plenderleith and Torraca 1968:243; Plenderleith and Werner 1971:198). Adjustable vertical anode sheets are hung on either side of the vat, and another anode is laid along the bottom of the vat. This is sometimes referred to as the 'sandwich setup.' The oxygen evolved off the bottom anode sheet ensures that the solution is continually mixed, preventing chlorides from concentrating along the bottom of the vat. The increased circulation also helps to maintain the anodes in a passive state by preventing the formation of strongly oxidizing, acidic hypochlorite from forming on them.

Sandwich setups have all of the disadvantages discussed above. An additional disadvantage is that the proximity of the object to the negative terminal connection is a factor in the current flow. The closer an artifact is to the negative terminal the more current it receives. By regularly repositioning the artifacts, each object will receive an average current for the duration of the treatment. Sandwich setups have the advantage of making it possible to process a number of specimens on a single power supply in one vat. This consideration is important when limited facilities are available to conserve a considerable amount of small articles.

When a metal vat is used as the anode, it is often advantageous to have auxiliary, adjustable expanded steel anode sheets hung along the sides and across the top of the vat to adjust for variations in the size of artifacts. When plastic vats are used, it is desirable to have a bottom anode because the oxygen evolved from the bottom ensures that the solution is continually mixed, thus preventing the tendency for molecular chlorides and hypochlorite to concentrate along the bottom or at the anodes.

Other refinements can be made to improve the sandwich setup for electrolytic cleaning. Most descriptions (Plenderleith 1956:195) recommend that three brass rods be suspended across the top of the vat, with vertical sheets of steel hung by copper wire from the side rods, and artifacts suspended by copper wire from the center rod. It is much simpler to use expanded steel sheets or stainless steel sheets, which extend up to the top of the vat and are bent and extended over the ends of the vat. This eliminates the two brass anode bars and the copper suspension wires, which would in any case go into anodic dissolution and plate the cathode. The anode sheets are attached to the positive terminal with a Mueller clip. Since the copper wire attached to the clip is out of the solution, it does not go into anodic dissolution and disconnect from the clip, which is a common occurrence on anode connections submerged in the electrolyte.

Suspending artifacts from the cathode rod with copper wire is not effective. It is difficult to maintain a good contact between artifact and the cathode; at some point, electrical contact will inevitably be lost. It is more desirable (at least for most small objects) to attach artifacts to the cathode rod with double-ended clips. Such clips can be purchased or made by bolting the ends of two clips together. The clips apply a constant pressure and ensure maintenance of a secure contact on the cathode rod and on the artifacts. The clips also facilitate attaching and removing artifacts without unnecessary difficulty.

In all Type 3 electrolytic setups, the conservator is unable to regulate the electrode potential or the current density to each artifact, which lessens the possibility of reducing the appropriate corrosion compounds back to a metallic state. It is also impossible to monitor the chloride loss from any one artifact in Type 3 setups.

There are also a number of setups that involve multiple artifacts, each in an individual cell, but all connected to a single power supply. In Figure 10B.1E, an example is shown using a mild steel or stainless steel vat with compartments; the vat is connected to the positive terminal and serves as the anode. An artifact can be placed in each of the compartments. When more than one compartment is used, the current to each artifact cannot be controlled, but the chloride level of the electrolyte in each of the compartments can be monitored. This approach takes advantage of limited facilities by using one DC power supply for several artifacts. It also enables the conservator to determine exactly when each object is cleansed of chlorides, keeping the length of treatment of each artifact to a minimum. When an artifact is completed, its compartment can be taken down and set up again without disturbing the electrolytic treatment in other cells.

Many conservation laboratories use an electrolytic setup in which a single power supply is connected to a control panel consisting of a number of amperage gauges and rheostats. Each artifact is placed in a separate vat and the current that runs from the power supply to each vat is regulated by a rheostat wired into the line. Each artifact or vat has a gauge that indicates the incoming amperage. This particular setup can be used to regulate the current to a number of artifacts from a single power supply; it has no particular disadvantages as long as the power supply has an amperage output sufficient for processing the number of artifacts that are being treated.

The two electrolytes commonly used in conservation for treatment of iron objects are alkaline solutions of sodium carbonate (Na2CO3) and sodium hydroxide (NaOH). Each electrolyte has its advantages and disadvantages and the conservator must know how to choose the electrolyte best suited for the object to be treated. Alkalies (and acids) used in conservation should be concentrated enough to do the required job but no stronger than necessary. This avoids over-cleaning the specimen and helps keep operating costs as low as possible.

In most laboratories that treat iron recovered from marine sites, a 2-5 percent solution of sodium hydroxide is the standard electrolyte used in electrolytic cleaning; it is the only electrolyte used when the objective is to maximize the reduction of ferrous corrosion products. Sodium hydroxide is more soluble in solution than sodium carbonate; unlike sodium carbonate, however, it will generate extreme heat when mixed in concentrated solutions. With its higher pH (12.9), sodium hydroxide is much more caustic than sodium carbonate and presents a potential hazard to those working with it. Caution must be exercised, and adequate safety equipment such as gloves, eye shields, eye washes, and safety showers, should be available.

If maximum reduction of corrosion products is not the objective of electrolytic treatment, a 5 percent sodium carbonate electrolyte with a pH of 11.5 will suffice for the cleaning of most iron artifacts.. In terms of safety, it is much less caustic than sodium hydroxide and much safer to handle. It is less conductive than sodium hydroxide, however, and must be used in stronger concentrations (5-10 percent vs. 2-5 percent). The more concentrated solutions of sodium carbonate are only slightly less expensive than their weaker sodium hydroxide counterparts, and sodium carbonate is usually easier to obtain from chemical supply houses than sodium hydroxide.

In experiments comparing artifacts treated in 5 percent sodium carbonate mixed in tap and de-ionized water and 2 percent sodium hydroxide mixed in tap and de-ionized water, chlorides diffused out of the artifacts and reached higher Cl- concentrations in sodium carbonate electrolytes more quickly than comparable artifacts in sodium hydroxide electrolytes (Locke n.d.).

The major problem encountered with sodium carbonate electrolytes is the precipitation of insoluble carbonate on artifacts during electrolytic cleaning (see below). Another problem commonly encountered with sodium carbonate electrolytes is that the pH and conductivity of the solution are inadequate in keeping mild steel anodes passive in the presence of high chloride levels. In sodium carbonate electrolytes, (OH)- ions discharge as oxygen at the anode more readily than the dissociation products of carbonate ions (CO3),2-. The anode becomes acidic by the accumulation of hydrogen from the oxygen evolution reaction; therefore, anodic dissolution is more prevalent with sodium carbonate than when sodium hydroxide, with its surplus of hydroxyl ions, is used. To ensure passivity in sodium carbonate electrolytes, mild steel anodes must be cleaned and replaced more often than in sodium hydroxide electrolytes, especially if the electrolyte is not circulated. It also takes longer to rinse out all residue of a sodium carbonate electrolyte from an artifact than to it does to rinse out all residue of a sodium hydroxide electrolyte.

The most important disadvantage of Na2CO3 relates to its pH and cathode reduction potentials. Theoretically, more efficient reduction of ferrous corrosion compounds is possible using 2-5 percent sodium hydroxide with a pH of 12.9 than 5 percent sodium carbonate with a pH of 11.5. (This is discussed in detail in the section on electrode potentials.) In general, sodium carbonate can be used as an electrolyte if reduction is not the objective; when treating metals from marine sites, however, a sodium hydroxide solution is the preferred electrolyte.

The cathodic precipitation of insoluble carbonate onto artifacts that are being electrolytically cleaned is the major problem encountered with sodium carbonate electrolytes. The carbonate precipitates from the electrolyte; this precipitation is more likely to occur at high current densities and when tap water containing high levels of carbonate is used to prepare the electrolyte. Carbonate precipitation in sodium hydroxide electrolytes is very rare and appears to be directly associated with the use of tap water that contains high levels of carbonate.

Artifacts recovered from salt water that are encrusted with calcium carbonate, magnesium hydroxide, and other minerals can provide the necessary elements to react with the carbonate in the electrolyte to form insoluble carbonates, such as calcium or magnesium carbonate. Once the surface of an artifact is plated with a carbonate deposit, chlorides may remain trapped beneath the surface. This will result in inaccurate chloride level readings in the electrolyte and lead the conservator to incorrectly assume that the removal of chlorides from the artifact is complete.

As a precautionary measure, gluconic acid, sodium gluconate, or sodium glucoheptanate in the amount of 2 percent of the NaOH or Na2CO3 in the electrolyte can be added as a sequestering agent to prevent the precipitation of insoluble calcium carbonates onto the artifacts. Gluconic acid and sodium gluconate require an excess of hydroxide in the electrolyte to work effectively; the lower pH of 5 percent sodium carbonate (pH 11.5) electrolytes will inhibit the action of these sequesterants. For both sodium carbonate and sodium hydroxide electrolytes, 50 percent aqueous sodium glucoheptonate has been found to be the superior sequesterant. In addition to preventing undesirable carbonate deposits on the cathode, the gluconate ions will sequester dissolved ferric ions, which would ordinarily precipitate as ferric hydroxide or oxide on steel anodes; gluconate ions will also prevent the formation of rust on artifacts during the rinsing and drying process.

If the local tap water is suspected of contributing to the formation of carbonate precipitate, de-ionized water should be used in the electrolyte. If carbonate precipitate is not observed, there is no reason to go to the trouble of adding a sequesterant to the electrolyte.

If carbonate does precipitate on an object, it is usually impossible to brush off or remove by electrolysis; rather, it can be eliminated by soaking the artifact for several days in a 5 percent solution of sodium sesquicarbonate or sodium hexametaphosphate. Sodium sesquicarbonate complexes with the insoluble calcium or magnesium salts to form soluble salts in the same manner as sodium hexametaphosphates (Plenderleith and Werner 1971:253).

It is generally recommended that only distilled or de-ionized water be used in electrolyte solutions. The exception to this recommendation occurs when the objects to be cleaned are heavily contaminated with chlorides. It is more economical to use tap water in the electrolyte until the chloride level approximates the level of the local tap water; once this level has been reached, the tap water is replaced with deionized water. The use of tap water in the first electrolytic baths may actually reduce the electrolysis time. Preliminary experiments indicate that chlorides are removed more rapidly when tap, rather than de-ionized, water is used in sodium hydroxide and sodium carbonate electrolytes.

When iron objects recovered from the sea are cleaned there is a very large buildup of chlorides in the electrolyte for the first few baths. If the metal vat that contains the electrolyte is also used as the anode, high chloride levels (2,000 to 18,000 ppm) and lack of external circulation can cause the vat to corrode extensively through anodic dissolution. Anodic dissolution is best prevented by using a 5 percent NaOH electrolyte with its surplus of hydroxyl ions; after the chloride levels drop below 1000 ppm, 2 percent NaOH can be safely used. Anodic dissolution is more frequently encountered when using a 5 percent sodium carbonate electrolyte and should not be used in vats that are not easily replaced.

The current density used in electrolytic reduction is expressed as the number of amperes per unit of artifact surface area that is introduced into the electrolytic cell by an external DC power supply, such as one ampere per square centimeter (1 amp/cm2). Current density ranging from 0.001 to 1 amp/cm2 have been proposed for use in electrolytic cleaning (Plenderleith 1956:195; Plenderleith and Torraca 1968:242; Plenderleith and Werner 1971:198; Pearson 1972a:12; Townsend 1972:252), but guides to the application of specific current densities are seldom given.

The irregular shape of many metal artifacts may make it difficult to determine their surface areas. Although it is seldom stated in conservation literature, the total surface area of the cathode is calculated only when an anode is form-fitted around the cathode and completely encompasses the artifact; otherwise, only the surface area of the object that is exposed opposite the anode is calculated. If the artifact to be treated is solid, well-consolidated metal, or if it has only ferric corrosion compounds, the applied current density is not critical. If the object has ferrous corrosion compounds, an initial high current density that results in the steady, vigorous evolution of hydrogen will quickly flake off the corrosion layer and significantly alter the shape of the specimen. In addition, an initial high current density can disfigure the surface of iron and also seal off the object, preventing the removal of deeply seated chlorides. These problems can be avoided by using a low current density in the early stages of electrolytic cleaning. More importantly, it is possible to reclaim enough of the metal through a reduction process that will consolidate the metal oxide interface, preserving a closer approximation of the object's actual dimensions. This is true even of wrought-iron objects whose corrosion layers are often non-adherent.

In Hamilton (1976:41), the following current densities and objectives are proposed. These current densities are recommended only as rough guidelines for treating iron artifacts recovered from marine environments. The densities were established by measuring the electrode potential established first on the surface of a rectangular bar of steel and then verified on a number of small artifacts.

  1. Low current densities (0.001-0.005 amp/cm2) facilitate the reduction range of ferrous corrosion compounds.
  2. Moderate current densities (0.05 amp/cm2) attempt to approximate the optimum conditions for chloride removal without undue evolution of hydrogen.
  3. High current densities (0.1 amp/cm2) encourage the vigorous evolution of hydrogen for mechanical cleaning.

If iron reduction is the objective, the specimen should be started off at a low current density. It is followed by a medium current density for the long chloride removal stage. A continuation of low current density would only lengthen the process of chloride removal, while the vigorous production and evolution of hydrogen at high current density interferes with the efficient removal of chloride from the metal and its corrosion products. At both low and medium current density, the marine encrustation and some corrosion products are slowly removed from the object by the mechanical action of the evolved hydrogen. For maximum efficiency, however, the artifact should be finished at a high current density. This ensures the complete removal of any remaining marine encrustation and loose corrosion layers, as well as any remaining chlorides. When high current densities are used only in the last stage of electrolysis, the metal corrosion products that are capable of being reduced have already been reduced, and it is less likely that they will be removed by vigorous hydrogen evolution. While the current density recommendations serve as useful guidelines, they cannot be adhered to rigidly. The conservator should take into consideration the individual peculiarities of the objects being treated and the facilities available to treat the objects. In the case of very large objects, for example, the above current density recommendations may require current in excess of the maximum capacity of an available or practical power supply.

For more direct control over the reduction of metallic corrosion compounds, North (1987:226) recommends that an artifact be placed in electrolysis and the voltage of the power supply slowly increased over the period of a week. The conservator graphs the progression of the treatment by plotting amperage along the x-axis and voltage along the y-axis. The voltage at which the amperage begins to rapidly increase is the hydrogen evolution voltage; this is the voltage at which to conduct electrolysis in order to optimize reduction. By using this technique, some of the difficulties of using current densities are eliminated, and it use is recommended.

In practice, neither measuring current density nor establishing hydrogen evolution voltage is necessary for most artifacts; merely observing the evolution of hydrogen from the artifact is often sufficient. In this technique, as the voltage or amperage is slowly increased, a few bubbles of hydrogen are observed evolving from the surface of the artifact. (In theory, metal reduction does not require any evolution of hydrogen, but the irregular evolution of hydrogen from the surface of the metal is a visual indicator that the current is flowing.) The artifact is treated at this level of low hydrogen evolution to reduce as many of the corrosion products as possible. After a short time, an increase in the evolution of hydrogen may be observed as the resistance of the artifact is broken down and some of the corrosion products in the metal are reduced; if this occurs, the current is decreased to the point at which the irregular evolution of hydrogen is again observed. After most of the desired reduction has been achieved, the current is increased so that there is a steady, yet not vigorous, evolution of hydrogen from the surface of the artifact. This is necessary to remove the maximum amount of chlorides within a reasonable length of time. Vigorous hydrogen evolution indicates a high current density that will mechanically clean the artifact. (As stated above, an artifact should not be mechanically cleaned at high current densities until all desired corrosion products have been reduced and chlorides removed.) In practice, this is how most experienced conservators determine the current density at which to treat iron artifacts as well as artifacts of other metals.

Current Density
Fig. 10B. Current Density

The applied current density will determine the electrode potential established between the electrodes and the electrolyte, as well as the rate of hydrogen evolution at the cathode. For reduction of metallic corrosion compounds, the establishment of a specific electrode potential on the cathode is essential. Since there is not a corresponding relationship between a given current density and a electrode potential on differentially corroded iron artifacts and different iron alloys, current densities and hydrogen evolution voltages alone are not precise enough tools for the treatment of very select artifacts. In such cases, it may be necessary to actually measure the electrode potential established at the surface of the artifact being treated. An understanding of electrode potentials also clearly demonstrates why one electrolyte is preferred over another when maximum reduction is the objective. This issue will be further discussed below.

In the case of iron, most of the corrosion products, if they can be reduced, are reduced to magnetite. Unlike non-ferrous metals, it has not been shown that any of the iron corrosion products actually reduce back to metal.

Control over the electrode potential at the surface of the cathode is the main objective and advantage of electrolytic reduction of metal artifacts. The normal electrode potential, or relative activity, of each metal is ranked in an electromotive series based upon a normal hydrogen electrode (Table 9.1). The hydrogen electrode is given an arbitrary potential of zero, and all metals are negative or positive to it. The normal electrode potential of a metal represents the electromotive force (EMF) required to balance the cell formed by a particular metal electrode immersed in a solution of its salt of normal cation activity and a hydrogen electrode (Evans 1963:231).

The potential series of the metals represent their equilibrium values at a pH of 0. These potentials change with pH value. In order for a metal to be corroded anodically, an electrode potential more positive than the equilibrium value of the metal must be established; a more negative potential must be created for cathodic reduction. Pronounced cathodic reduction of iron occurs only when the electrode potential is considerably more negative than the equilibrium value of the iron. Dissolution of the anode becomes significant when the electrode potential is made considerably more positive.

The electrode potentials of metals are altered if the [ionic] activity (effective concentration) is not normal. Clearly, if the solution is diluted, the passage from the ionic to the metallic state will be slower, whereas the passage from the metal into solution remains unaltered. Thus the balance will be upset, but a fresh balance may be obtained at a more negative potential. This means that at 18°C, if activity and concentration are regarded as identical, every ten-fold dilution (say from N to N/10, or from N/10 to N/100) will shift the potential in a negative direction by about 0.058 volts for a monovalent ion, or 0.029 volts for a divalent ion. The theoretical shift of potential is approximately realized in the case of the more noble metals; but on some of the less noble metals [iron] the potential actually measured is found to be almost independent of the concentration of the metallic salt in the original solution; this occurs if the metal is capable of reacting with the solution, so that the concentration in the layer next to the metal becomes different from that in the body of the liquid (Evans 1963:234). (See Figure 10B.2.)

The potential of iron immersed in an aqueous solution originally free from iron ions depends on the hydrogen-ion concentration, becoming steadily more negative as the pH value rises (Evans 1963:235).

The electromotive series in Table 9.1 indicates that the reduction reaction of ferrous ions, Fe2+ ⇔ Fe - 2e, occurs at -0.409 volts to -0.440 volts at a pH of 0. The reduction potential increases -0.029 volts for each pH increase. The reduction of ferric ions, Fe3+ ⇔ Fe - 3e, occurs at -0.036 volts and increases -0.019 volts per pH increase. The hydrogen discharge potential, 2H+ H2 -2e, occurs at a voltage potential of 0 and increases -0.058 volts per unit of pH increase. The reduction potential of ferrous, ferric, and hydrogen ions at different pHs at a standard temperature of 20°C are graphically presented in Figure 10B.3. This graph makes it possible to determine the theoretical electrode potential voltage necessary for reduction of ferrous corrosion products in an electrolyte of known pH and ion content. It is thermodynamically impossible to reduce ferric oxide and other ferric iron corrosion compounds in an aqueous solution; this point is also graphically depicted in Figure 10B.3. The electrode potential for the reduction of ferric corrosion products is well out of the range possible for electrolytic reduction in an aqueous alkaline electrolyte with an external EMF.

Figure 10B.3. Reduction potentials of ferric, ferrous, and hydrogen ions at different pHs and standard temperature of 20°C. The pH range of a 5 percent Na2CO3 and a 2 percent NaOH electrolyte are shaded.

Electrode potentials are affected by temperature, pH value and electrolyte composition. In the conservation of iron pH is the most important factor. There are several ways to measure the pH of the electrolyte but no simple way of determining the pH at the surface of an electrode; unfortunately, this is the pH reading of concern. It is known that the pH of the catholyte (the solution at the surface of the cathode where hydrogen is evolved) is higher than that of the remaining electrolyte. In order to arrive at a reasonable estimation of the prevailing pH at the cathode, the pH of the electrolyte is measured and a maximum increase of one pH level is assumed; this results in a maximum pH of 14 for sodium hydroxide and 12.5 for sodium carbonate. This approach is imprecise, but there is no direct method by which to measure the pH of the catholyte. These adjusted pH ranges are shown in Figures 10B.3, 10B.4 and10B.5.

Figure 10B.4. Reduction potentials at the surface of an iron cathode during electrolysis in a 2 percent NaOH electrolyte at different current densities over an 11-day test period.

Figure 10B.5. Comparison of the reduction electrode potential on the surface of the same iron cathode in a 5 percent Na2CO3 and a 2 percent NaOH electrolyte.

The electrode potential at the surface of the cathode is a critical factor in the reduction of corrosion compounds during iron electrolysis. As the hydrogen discharge potential of the cell is reached, hydrogen ions are formed which reduce adjacent ferrous compounds. (At more negative electrode potentials, molecular hydrogen will form so rapidly that the molecules will combine with each other and evolve as hydrogen gas; this evolution of hydrogen acts as a mechanical cleaning agent that physically removes any unconsolidated corrosion layers.) The theoretical point at which the maximum reduction and/or consolidation of the ferrous corrosion compounds is achieved is at the intersection of the hydrogen discharge line and the ferrous ion reduction line or immediately to the left of this intersection (see Figure 10B.3). Using the formula Fe2+ + 2e ⇔ Fe and substituting values for the Fe2+ reduction potential and the correction for pH [(-0.409)( -0.029)(pH)], this intersection appears at a pH of 14.1 and an electrode potential of -0.82.

Figure 10B.3 gives the measured electrode potentials of an artifact in an electrolyte of 2 percent sodium hydroxide with a pH of 12.9 at different current densities over an 11-day period. The initial test was taken before the surface of the artifact was thoroughly saturated with hydrogen, the current resistance of the electrolyte had dropped, and the resistance of the artifact was broken down. After a period of 20 minutes, and throughout the remainder of the 11-day test period, the electrode potentials fell within a rather narrow range. Using the data presented in Figure 10B.3, the potential range for ferrous ion reduction at a pH range of 13 to 14 can be determined as -0.79 to -0.85 volts. In Figure 10B.4, this potential range is achieved at a current density below and up to 0.005 amps/cm2, which is barely within the iron reduction range of NaOH. Figure 10B.5 compares the electrode potentials of the same artifact in 2 percent NaOH and 5% Na2CO3; the pH of sodium carbonate is not high enough to establish the desired theoretical electrode potentials for the reduction of ferrous compounds. On theoretical grounds, a sodium hydroxide electrolyte is clearly superior when reduction of the ferrous corrosion products is the objective.

After electrolysis has begun, several hours to a full day are required for the cathode to adjust to an equilibrium state. At the metal surface, a metal/hydrogen bond is established, and the surface becomes saturated with hydrogen, followed by the evolution of hydrogen. Until the metal surface becomes saturated with hydrogen and establishes equilibrium, the electrode potential will remain in flux. After the electrode potential on an artifact has been established, daily adjustment of the current may be necessary to maintain the potential as the current resistance of the electrolyte drops with the addition of dissolved salts, and the resistance of the metal and its corrosion products is altered. The procedure for measuring electrode potentials is described in Hamilton (1976:103-105). Before adjusting the current back to the predetermined potential, however, it must be certain that the shift in potential is due to changes in the cathode and not to the transport system (i.e., changes in the electrolyte due to lack of circulation). In Figure 10B.4, the six series of tests were taken over an 11-day period without circulating the electrolyte. On the last day, the electrolyte was circulated and the test was run again. There are negligible differences in the electrode potential readings of the last two tests, except for the potentials at high current densities. The transport system is still adequate, especially for the electrode reduction potentials of ferrous compounds. If any changes occur in the transport system, it will become visually apparent at the anode, where acidic areas may form. In these areas chloride ions react with the positively charged anode to form hypochlorite ions (ClO-) or hypochlorous acid (HClO), which will oxidize the anode. Adverse reactions are easily detected at the anode, and when this occurs, the cathode is also affected, although it is not as apparent or as detrimental to the operation of the electrolytic cell. The potential at both electrodes, however, is altered. Should a change in the potential occur, the electrolyte should be stirred or circulated and the measurements taken again. If the electrode potential remains the same, then the amperage may be adjusted back to re-establish the optimum range.

From the standpoint of efficiency, electrode potential measurements are necessary for precise control and quality processing of very select artifacts. The areas of advanced or retarded corrosion on the surface of the artifact will be reflected in the electrode potentials, making it possible for the conservator to identify those areas requiring more extensive reduction and to measure the progress of the electrolyte reduction. The data acquired from electrode potential measurements have been used to arrive at realistic current densities for the bulk of the iron artifacts that require treatment but do not warrant the close supervision or the time required for taking electrode potential measurements. Still, there is the difficulty of arriving at the most effective current densities for given artifacts, especially irregularly shaped artifacts and differentially corroded artifacts. For these reasons, using either hydrogen evolution potential voltage or merely visually observing the rate of hydrogen evolution is sufficient for the bulk of iron artifacts.

All electrochemical reduction cleaning processes produce irritating caustic vapors and hydrogen gas, which must be properly exhausted from any room or building for the safety and comfort of personnel. Expensive fume hoods or sealed rooms with exhaust systems are required; alternatively, the electrolytic vats can be placed outside. Most laboratories use variations of all three methods. The smaller artifacts are cleaned in the main laboratory under fume hoods. A few large artifacts are cleaned in a ventilated room in a separate building. Other large artifacts and some small ones are electrolytically cleaned outside, in vats made of mild steel or plastic.

Expensive air exhaust systems are not necessary when electrolysis is performed outdoors. The solutions are changed frequently enough so that any air-blown dirt or sand in the electrolyte will present no problems. Rain is of no concern, unless it is heavy enough to substantially dilute the solution. This problem can be easily determined with a hydrometer to test the specific gravity, and remedied by adding more sodium hydroxide or sodium carbonate. In many cases, rain will only decrease the amount of de-ionized water that needs to be added to the vats daily to compensate for evaporation, caustic vapors, and electrolytic reduction.

If the area is fairly isolated (i.e., safe from human traffic) and has adequate utilities, it is strongly recommended that outdoor electrolytic reduction be performed. The power supplies must be kept out of the weather, but sheds over the vats are optional. Although lower temperatures do not significantly decrease chemical activity, there may be an IR drop in the electrolyte that will increase the cell voltage. All electrolytes should be kept from freezing.

The following observations are directed primarily towards large iron objects but are not restricted to them. Recommendations for improving electrolytic cleaning procedures and reducing electrolysis time are offered.

Although conservators often avoid cleaning large iron objects electrolytically (because of their size), there are no insurmountable problems. In order to better understand the factors that lead to success or failure in electrolytic cleaning, several case histories are reviewed.

The earliest attempt to electrolytically clean a large, seabed iron artifact was by Lieutenant Nielsen of Norway. The specimen was a wrought-iron gun recovered in 1942 from a ship dating to the 15th century (Eriksen and Thegel 1966:100-102). The conservation was attempted under the adverse conditions of World War II, with inadequate facilities and supplies. The attempt was unsuccessful for a number of reasons, the most important being the following:

  1. Failure to remove the wooden undercarriage shielded a large portion of the cannon from the electrolyte and prevented an even distribution of the current.
  2. The electrolytic treatment was too brief for complete chloride removal, lasting a total of 69 discontinuous hours spread over a 384-hour period.
  3. The placement of the steel anode sheet along only one side of the vat failed to ensure an even distribution of current.
  4. After the brief electrolytic treatment, the wooden undercarriage was swabbed with hydrochloric acid to neutralize the 1 percent sodium hydroxide used as the electrolyte. This acid introduced additional chlorides to attack the iron.

Ora Patoharju (1964, 1973) reports one of the first successful attempts to electrolytically clean a large cast-iron cannon. In 1963, two guns from a 1790 shipwreck were treated. One had been recovered many years before treatment and had been allowed to dry out; the second cannon had recently been recovered and had been kept wet. Both served as the cathode with three stainless steel sheet anodes in a 10 percent Na2CO3 electrolyte. They were run at a rate of 20 amperes at 4 volts (current density ca. 0.027 amps/cm2) for one month and a rate of 150 amperes at 4 volts (current density ca. 0.2 amps/cm2) for five months.

Not enough details on all aspects of the electrolytic treatment were provided in order to properly evaluate the methods employed. After being displayed outdoors for two years, however, the 'wet' gun was stable and its treatment judged to be successful, although the dry gun continued to deteriorate as rapidly as before electrolysis. The most likely reasons for the successful treatment of one gun is that it was kept wet, preventing further corrosion, and that the six months of electrolysis ensured the complete removal of the chlorides in the metal. The preliminary conclusion of Patoharju is that cast-iron objects coming directly from the sea should be kept wet up to the point at which they are immersed in the electrolyte. If the object is allowed to dry out before treatment, the specimen will undergo further oxidation, and any conservation attempt may be futile.

In another example, a couple of large breechblocks had been successfully treated with electrolysis, with the majority of the corrosion layer left intact on the specimens. A month after completion of the treatment, it was decided to put the breechblocks back into treatment in order to darken the surfaces. The wax that coated the breechblocks was melted off in an oven, which most likely caused increased conversions of FeO to Fe2O3 on the artifacts before they were put back into electrolysis. In this case, and that of Patoharju's 'dry' gun, the corrosion layer completely detached and left behind a badly disfigured surface.

The above examples suggest that if there are thick layers of corrosion, the treatment must be successful the first time. Wrought-iron corrosion layers are non-adherent; after they have been removed of water during the drying and sealant steps, these layers apparently undergo additional oxidation and may also lose the electrical couple with the metal core. During subsequent electrolysis, hydrogen will evolve at the surface of the metal and slough off the corrosion layer. This problem may possibly be alleviated by soaking the object in electrolyte for a long time prior to any further electrolysis in order to re-establish the electrical couple.

Pearson (1972a, 1972b) discusses the conservation of six cast-iron cannons and other relics jettisoned in 1770 by Captain Cook on Endeavor Reef, off of the Australian coast. Upon arrival at the laboratory, the cannons were stored in 2 percent solution of sodium hydroxide, and each was placed in an epoxy-coated wooden cradle specifically constructed to support it throughout the conservation process. The coral encrustation was mechanically removed with hammers, and the cannon surfaces were carefully scrubbed with water to remove the loose black corrosion products. The three cannons and their support cradles were simultaneously treated in separate fiberglass-lined wooden tanks with a 2 percent NaOH electrolyte. The vats were hooked up in series on one power supply, analogous to the Type 4 setup described earlier. Each cannon was set up as the cathode and two anodes, a single mild steel sheet along one side of the vat, and a steel rod in the bore, were used. A current density of 10 amps/m2 (0.001 amps/cm2) was applied. Each week, the third cannon was removed from its vat and the electrolyte discarded and replaced with fresh electrolyte. Cannon No. 1 was then placed in this vat; Cannon No. 2 was placed in the vat from which No. 1 had been removed; and Cannon No. 3 was put in the vat formerly occupied by Cannon No. 2. With this method, the first cannon, the one in electrolysis the longest, was always placed in the fresh electrolyte. This rotation was continued weekly until analysis showed that there was no increase for one week in the chloride content of the bath containing the first cannon. This method required six to eight weeks of electrolysis for each cannon.

A current density of 0.001 amps/cm2 was chosen by Pearson because experiments proved it to be the optimum current density value for the removal of chlorides. Pearson found that higher values blistered the graphitic surface layers of the cast iron, and lower values only prolonged the time required to remove chlorides. More importantly, this current density is within the range that is most efficient for the reduction of ferrous iron compounds. A series of experiments (Hamilton 1976:40-46) on wrought iron indicate that 0.001 amps/cm2 to 0.005 amps/cm2 is theoretically most efficient for metal reduction (as is shown in Figures 10B.6 and 10B.7); 0.05 amps/cm2 to 0.1 amps/cm2 is most efficient for chloride removal; and the vigorous evolution of hydrogen produced by a density of 0.1 amps/cm2 and above is used for mechanical cleaning.


Fig 10B.6 and Fig 10B.7

When Pearson (1972a) cleaned Captain Cook's cannons, he placed a single mild steel anode along one side of the vat and used immovable wooden support cradles on the cannons. The placement of the anode failed to create an even current density, and the cradles shielded two areas from what current flow was present. This shielding interfered with the removal of the chlorides and the reduction process. Pearson (personal communication) reports that minor corrosion occurred at these two support points because the cannons were not rotated sufficiently during the treatment. Similar occurrences have occurred even on small surface areas covered with plastic identification tape.

From these examples, it is clear that no portion of the artifact can be covered throughout electrolysis; even a small piece of narrow tape can shield the area beneath it enough to prevent complete removal of the corrosive chlorides. To avoid this problem, cannons should be rotated frequently, and the position of movable supports should be shifted each time the electrolyte is changed. When plastic tape is placed on artifacts for purposes of identification, the position of the tape should be moved at least once during electrolysis.

The importance of using form-fitted anodes in electrolysis should be emphasized. Such anodes, positioned to ensure an even current density over all the surface of the artifact, improves the overall efficiency of the electrolytic process. An exact distance does not need to be maintained between the artifact and the anode at every point; this is nearly impossible for irregularly shaped artifacts. Rather, the anode should be bent and shaped so that is somewhat evenly distanced from the artifact at all points.

North (1987:225), citing Hamilton (1976), states that the use of form-fitted anodes has been suggested, but that they are unnecessary because a satisfactory current distribution will be achieved as long as the artifact-anode distance is between 20-80 cm. However, he further states that this recommendation does not apply when a rod is used as the anode in the barrel of a cannon, as the current will be concentrated in the cannon bore due the small distance between the rod and the cannon. In addition, while discussing the establishment of hydrogen evolution potential voltage, North (1987:226) notes that hydrogen gas evolution is coming from only one area of the artifact may be due to the fact that the anode is too close to the artifact at that point. This clearly demonstrates that maintaining a relatively even distance between the artifact and the anode is essential for proper electrolytic cleaning.

Even if form-fitted anodes are not used for every electrolytic treatment (artifacts are frequently placed between two anodes, particularly when processing a large amount of artifacts at a single time), they are highly recommended. They will maximize iron reduction and chloride removal and will cut down on the electrolysis time.

Ora Patoharju (1964) reported an electrolysis time of six months to clean two cannons recovered from a 1790 wreck. In general, wrought-iron objects that approach the size of a cannon require between 6-12 months of electrolysis. As an example of the length of time required to treat large wrought-iron objects, one 198-cm long fully armed swivel cannon required 20 days to remove all of the encrustation and to dismantle the component parts; 251 days of electrolytic cleaning at 20 to 50 amps; 7 days of rinsing in several changes of alternate boiling and cold de-ionized water; 15 days of dehydration in alcohol; and 1.5 days submerged in molten microcrystalline wax. Three hooped barrel cannons processed together in one vat required 480 days of electrolysis, 3 months of rinsing (the rinsing period was prolonged while awaiting delivery of microcrystalline wax), 5 days of dehydration in alcohol, and 5 days submerged in microcrystalline wax. Even small wrought-iron artifacts, such as spikes, often require 60-90 days of electrolysis. If an object is kept in electrolysis for too short a time, the full potential of electrolysis is not achieved, and the time required to rinse the artifact is significantly increased.

Reducing Electrolysis Time
When artifacts are electrolytically cleaned in small vats with a low ratio of electrolyte to artifact volume, the length of electrolytic time is considerably extended. The larger the volume of electrolyte to artifact being cleaned, the shorter the period of electrolysis required. Two wrought-iron breechblocks that were cleaned in two small vats required 25 months of electrolysis. In addition, more frequent changes of the electrolyte will also considerably decrease the time required to electrolytically clean an artifact.

The stage at which the chloride (Cl-) level in the electrolyte is reduced to below 50 to 100 ppm requires the greatest amount of electrolytic cleaning time. Below 100 ppm Cl-, the artifact is assumed to be in the final stages of cleaning. There are at least two explanations for the long amount of time required to remove chlorides during electrolysis: (1) The electrolysis is carried out at high current densities which have a tendency to repress Cl- migration in preference for H2 evolution, (2) The Cl- concentration in the electrolysis is governed by the Donnan equilibrium theory.

The Donnan equilibrium theory pertains to the unequal distribution of ions on two sides of a membrane (Kunin 1958:14-16). Although no membrane exists between the artifact and the electrolyte, the interface between the solid and liquid phases may be considered a membrane. An exchange of Cl- ions continues until the concentration ratios are equal in both phases. The lower the Cl- ratio in the electrolyte, the more effective the diffusion of the Cl- ions from the artifact to the electrolyte. Using the Donnan equilibrium theory, it can be generalized that the rate of Cl- transfer from the artifact to the electrolyte is negated despite the electrolytic field when the Cl- content of the electrolyte reaches equilibrium with or is higher than the Cl- level in the artifact. Chloride tests can be misleading under equilibrium conditions. When the Cl- content in the electrolyte maintains a steady level in a controlled constant volume of electrolyte, it should be determined that either the chloride removal process has been completed, or that equilibrium conditions have occurred. Unless the electrolyte is renewed to lower the Cl- concentration, little or no further Cl- removal from the artifact can be achieved. Either determination can be easily validated by changing the electrolyte and monitoring the chloride levels for a few more days. Frequent changes of the electrolyte and use of as large a vat as possible ensures that the chloride level of the electrolyte is lower than the chloride level of the artifact. This will facilitate chloride removal and make the entire electrolytic process more efficient. For example, two similar wrought-iron anchors were placed in electrolysis, with the electrolyte in one anchor vat changed more frequently than the electrolyte in the other vat. The anchor that underwent more frequent changes of electrolyte required 6 months of electrolysis in contrast to 11 months for the other anchor. Three wrought-iron cannons required 16 months of electrolysis at 100-150 amps and 3 volts to remove the chlorides, whereas a second group of three hooped barrel cannons that underwent more frequent changes of electrolyte were completed in 11 months.

When sea-recovered artifacts are cleaned by electrolytic reduction, it is recommended that the artifact be started in an electrolyte of 2-5 percent NaOH in tap water. During this period, the reduction of ferrous compounds is attempted at low current density, and high levels of chloride are removed at medium current density. Electrolysis is continued in the tap water electrolyte until the chloride level of the electrolyte approximates the level found in the tap water. Tap water is then replaced by de-ionized water, and electrolysis is continued with 5 percent NaOH or lowered to 2 percent NaOH. A low current density is maintained until the electrolyte is again changed, whereupon a medium current density is used.

The higher pH of NaOH is preferable when the objective of electrolysis is the reduction and/or consolidation of ferrous corrosion products. If this is the case, NaOH should be used as the electrolyte. If the corrosion products are in a ferric state, there is no advantage in using NaOH, and Na2CO3 can be used from the beginning. This is applicable to many iron objects that are air-oxidized in archaeological sites, but this generalization should not be carried out too far. There will almost always be some ferrous compounds that can be successfully reduced.

The alkaline sulfite treatment was developed by North and Pearson (1975b) to stabilize marine cast iron, but it is also used on wrought iron. Bryce (1979:21) found that the treatment is effective on iron objects that are moderately to heavily corroded, but the objects still must have a metallic core present for the treatment to be effective; otherwise, the iron object will break apart during treatment. The procedure is as follows:

  1. 1. After mechanical cleaning, an iron object is immersed in a solution of 0.5 M (20 g per liter of water) of sodium hydroxide and 0.5 M (126 g per liter of water) of technical-grade sodium sulfite. A glass container should be filled with as much solution as possible and the object quickly placed in it to avoid any oxidation of the solution. The container is then sealed air tight, placed in an oven, and kept heated to a temperature of 60°C. The object is processed through several baths of the solution until chlorides are eliminated; this may take a week to several months and numerous baths. Tap water can be used for the first one or two baths, but de-ionized or distilled water should be used in the final baths. The solution does not attack any residual metal, so there is no danger of too many baths.
  2. When a marine iron object is immersed in this hot reducing solution, the iron corrosion compounds are converted to magnetite, and the chlorides are transferred to the solution and discarded with each bath change. The object comes out of the treatment with a very black surface coloration. Since the solution is strongly alkaline, contact with the skin should be avoided.

  3. Once the alkaline sulfite stabilization treatment is completed, the object is washed for one or more hours in several baths of de-ionized water and then placed in a 0.1 M solution of barium hydroxide (32 g per liter of water). Barium hydroxide is slightly poisonous, so contact with the skin should be avoided. If the object is intensely rinsed in several baths of de-ionized water following the alkaline sulfite stabilization, the barium hydroxide baths can be eliminated.
  4. The alkaline treatment has been very effective for conserving iron recovered from a marine environment. The main drawbacks of the treatment are that it must be carried out in an air-tight container, and the container should be kept heated.

A number of chemical cleaning treatments are used for iron artifacts recovered from non-marine environments that have negligible chlorides present. The most common chemicals used are oxalic acid, citric acid, phosphoric acid, ethylenediamine tetra-acetic acid (EDTA), and other complexing agents. The exclusive use of any of these chemical treatments may improve the appearance of an object, but they do not remove chlorides and hence cannot prevent subsequent corrosion. Therefore, they are not considered as conservation alternatives for treating iron recovered from salt water. Details concerning the use of these and other chemicals are described by Plenderleith and Werner (1971).

Two chemicals, phosphoric acid (and its derivatives in commercial rust removers) and tannin solutions, are often used to form a corrosion-resistant film of phosphate and tannates on the surface of treated iron pieces. The corrosion-resistant significance of phosphate and tannate films was first made apparent when iron articles recovered from an ancient Roman tannery in England were found to be in an excellent state of preservation (Farrer et al. 1953). Before either chemical can be used, however, the chlorides must be removed by electrolysis, alkaline sulfite treatment, or water diffusion.

The corrosion-resistant nature of tannate films on iron was investigated by Knowles and White (1958) and later by Pelikan (1966). In accelerated exposure tests, it was found that tannate films on iron were more corrosion-resistant and lasted twice as long as phosphate coatings. It is important that the right tannic acid be selected, as many are not effective. Argo (1981) provides a discussion of the requirements and benefits of tannic acid solutions. Tannic acid solutions (such as Baker reagent tannic acid, C76H52O46) with a pH of 2.5-3.0 provide good, weather-resistant tannate films. Solutions of hydrolysable tannins, such as extracts of chestnut, myroblans, or valonea, with a pH of 2 to 2.5, provide the most weather-resistant protection (Knowles and White 1958:16). If the tannic acid mixture has too high of a pH, phosphoric acid should be added to bring it down to a pH of 2.4. In most conservation laboratories, tannic acid solutions are a standard part of the conservation of all iron artifacts. Although a tannic acid coating is often the final step in iron conservation, it is recommended that an additional sealant, such as microcrystalline wax, be applied over the oxidized tannate film for maximum protection.

In the past, the Conservation Research Laboratory at Texas A&M University has used a 20 percent tannin solution (200 g tannin, 1 liter water, 150 ml. ethanol) on iron artifacts. Logan (1989) states that solutions of tannic acid stronger than 10 percent are much too concentrated and recommends that several coats of diluted 2-3 percent tannic acid (with the addition, if necessary, of sufficient concentrated phosphoric acid to achieve a pH of at least 2.4) be brushed on the object. She also cautions that different brands of tannic acid react differently and recommends BHD Chemical tannic acid as working consistently well. Recent treatments conducted by the Texas A&M Conservation Research Laboratory on iron artifacts recovered from a 17th-century French shipwreck have demonstrated that a 5 percent tannic acid solution permeates the corrosion layers of the artifacts better than a 10 percent solution. In general, a 2-10 percent solution of tannic acid is effective; the concentration of the solution should be determined by the ability of the solution to permeate the corrosion layers of the artifact to be treated.

Several coats of the tannic acid solution should be applied with a stiff brush to the surface of the artifact. A brushed-on film provides better protection than a dipped or sprayed application because the brushing ensures that the solution has access to the metal in areas of loose rust. Brush application also eliminates the polarization of cathodic areas by the formation of hydrogen (Pelikan 1966:112). Objects such as cannon balls have also been successfully treated by vacuum impregnating them with a tannin solution. The object is allowed to completely air oxidize between each application of the tannic acid; after the final application, the object should be allowed to dry for one to two days.

The tannin solution reacts with the iron or iron oxide to form a ferrous tannate, which oxidizes to a mechanically strong, compact, blue- to black-colored ferric tannate. In order to ensure a continuous tannate film, Knowles and White (1958) recommend that all iron oxide products be removed from the surface of the artifact, otherwise there is a possibility that corrosion may start at the junctures of the cathodic iron oxide and the tannate film. Good results can be achieved even if this recommendation is not followed; Pelikan (1966:110-111) found that tannin solutions react directly with the metal base and with the rust if the solution is sufficiently acidic ( pH 2 to 3). In addition to forming a corrosion-resistant film, tannin solutions can be used to impart an aesthetically pleasing black color to iron.

A phosphate film is formed on iron objects by impregnating them with a 20 percent solution of phosphoric acid  (H3PO4). Impregnation under a vacuum is recommended in order to ensure complete penetration of the acid into all porous areas of the metal. The acid complexes with the iron to form an inert film of ferric phosphate film on the surface of the metal. Reese-Jones (1972) describes the above procedure and reports that porous cast-iron cannon balls from a 1588 Spanish Armada shipwreck were successfully treated by this method after the chlorides were removed by water diffusion. Similar results can be achieved on wrought iron or steel.

Data reported by Pelikan (1966:112-113) indicate that a mixture of phosphoric acid and tannin solution can be used on badly rusted iron to appreciably improve the corrosion resistance of a phosphate film. One hundred milliliters of 80-85 percent phosphoric acid solution is added to the 20 percent tannin solution, and several coats are brushed onto the artifact. This is followed by at least four coats of the standard 20 percent tannin solution. Following the treatment of an object with the phosphoric acid-tannin solution, a final sealant should be applied to seal off the tannate or phosphate film. This solution does not result in the rich, dense black coloration that Baker tannic acid does by itself.

Regardless of whether tannic or phosphoric acid is used, it is highly recommended that a sealant, such as microcrystalline wax, be applied over the film formed on the object, The wax will provide a vapor barrier, which the film does not, and will also contribute some strength to corrosion layers on the metal.

Treating sea-recovered iron objects by heating them to a temperature of 850°C was first attempted in Denmark in 1955 (Eriksen and Thegel 1966). It was proposed that at this temperature, the chemically bound water associated with the ferrous chloride corrosion products in the porous matrix of the metal would be driven off, leaving behind inactive anhydrous ferrous chloride. Simply leaving chlorides in an anhydrous state, however, will not prevent subsequent corrosion. Both ferrous and ferric chloride are capable of absorbing water from the atmosphere and reinitiating the corrosion process unless a perfect air-tight, atmosphere-proof coating is applied. The success of annealing is more likely due to the sublimation of the chlorides. Experiments conducted by Pearson (1972a:25) indicate that a temperature of 850°C is well above the melting point of ferrous chloride, and both ferrous and ferric chloride sublimed strongly from a piece of test iron at 700°C. Therefore, it is probable that the sublimation of ferric and ferrous chlorides occurs at a temperature of 700°C and higher.

The apparent absence of corrosion during 20 years of outdoor exposure is testimony to the success of this method. However, there are considerable disadvantages to this method:

  1. There is no opportunity to preserve the iron oxide interface. The corrosion layers spall and crumble, obliterating any decoration or identification markings on the cannon and leave behind a badly disfigured surface.
  2. The remaining surface is oxidized to an unattractive red iron oxide color.
  3. The elevated temperatures alter the metallurgical microstructure of the metal, rendering the artifact useless for future metallurgical examinations.

Because of these shortcomings, the technique of annealing in an oxidizing atmosphere is not recommended.

Work by Barkman in connection with the preservation of the Wasa, the 1628 ship raised intact from the Stockholm Harbor in 1961 (Barkman and Franzen 1972), has demonstrated that annealing iron at a temperature of 1060°C in a reducing atmosphere successfully stabilizes and preserves it, returns it to a metallic state, and eliminates the corrosive chloride compounds by sublimation.

The iron to be treated is placed in a hydrogen furnace and heated in the presence of hydrogen gas. The temperature is slowly raised to 1060°C over a period of one week. At this elevated temperature, all of the moisture is driven off, and all of the chloride corrosion compounds are volatilized. The hydrogen reduces the iron corrosion compounds back to a metallic state and combines with the oxygen in the corrosion products, forming water, which is driven off by the heat. Over the years, the temperature to which the hydrogen oven is heated has been decreased. In the past, objects undergoing hydrogen reduction were placed in a special furnace with 100 percent dry hydrogen gas, or a mixture of hydrogen and nitrogen, initially heated to a temperature of 300°C and over a period of days taken to 1000°C (Barkman 1978:155-166).

Annealing in a reducing atmosphere is said to result in very little damage to the surface of the metal. More experiments are needed, however, to determine the extent of any undesirable surface alteration. From the standpoint of time and final results, the technique appears to be a satisfactory, efficient alternative for cleaning iron recovered from the sea.

The primary drawbacks to this technique are the expense of the equipment and the lack of hydrogen kilns big enough to treat large objects. At present, a few conservation laboratories use annealing to treat numerous small artifacts, such as cannon balls, with a minimum amount of hands-on handling. Another drawback is the large amount of hydrochloric acid produced when the chlorides are driven off from the artifacts; the hydrochloric acid will attack any exposed metal inside of the kilns or furnaces. The detrimental effect the HCl fumes on the metal hardware of the kilns or furnaces has discouraged some companies who own industrial furnaces from volunteering their use to conservators.

All annealing methods present the problem of the changes in the metallurgical characteristics in the metal when heated to high temperatures.

The loss of information by the treatment of totally rusted marine cast iron at 800°C will not be great and there seems to be little objection to the use of the hydrogen reduction process at 800°C for this purpose. The reduction of rust on wrought iron is a different matter. The main problem is knowing whether the residual metal contains, either intentionally or unintentionally, enough carbon to give useful information to the archaeometallurgist. If carbon is absent then treatment at 380°C is acceptable although some change will occur. The slag inclusions will suffer very little microscopic change and no macroscopic change. To ensure the removal of chlorides at 380°C the treatment time must exceed 60 hours (Tylecote and Black 1980:95).

As long as the conservator follows the recommendations cited above, the major objections to annealing in a reducing atmosphere of hydrogen are overcome. The main limiting factor is the high cost of this type of furnace and the inherent safety concerns and potential danger of heating hydrogen to very high temperatures. The treatment does result in stable, chloride-free artifacts.  

Hydrogen plasma reduction is a relatively new technique (see Patscheider and Vepek 1986). Iron artifacts, as well as those of copper and silver, are conserved by placing them in a quartz discharge tube surrounded by hydrogen gas under low pressure. The hydrogen gas is ionized into plasma by the introduction of high-frequency radio waves. Iron is in the center of the hydrogen plasma, and the magnetite and ferric oxide on the surface of the iron are converted to metallic iron. Because the treatment is carried out at a temperature of less than 400°C, there is no change in the metallic structure of the iron.

Although preliminary results have been encouraging, the primary disadvantage of this conservation technique is the high cost of the equipment and the small size of the artifacts that can be treated in all but the most expensive units.

In any archaeological excavation of a shipwreck, there will always be some artifacts that cannot be conserved by any the methods discussed above. Little or no metal may remain in the artifact; even if metal is present, any electrochemical or chemical treatment may considerably alter the form of the object. If the methods outlined above are not feasible for conserving a particular object, the conservator has three additional options: the chlorides can be removed by a process referred to as water diffusion; the article can be embedded in plastic; or the artifact can be cast (see the file on Casting and Molding).

The only way that an iron artifact recovered from the sea can be stabilized is to remove the chlorides from the metal. This is most easily accomplished by electrolytic reduction techniques, but if the artifact is very badly oxidized, and its overall form and dimensions are to remain intact, the only alternative is to remove the soluble chlorides by the much slower process of water diffusion (Oddy and Hughes 1970). The name accurately describes the process. The artifact is placed into a container filled with water, and the water is changed frequently, as the soluble chlorides diffuse out of the metal into the solution. The water should be changed weekly or as often as necessary as determined by a qualitative or quantitative chloride test. Because water diffusion does not remove chlorides from an artifact within in any accepted time frame, it should only be considered when attempting to conserve an artifact so badly corroded that there is the possibility that it may be destroyed if it were cleaned by electrolytic reduction or by the alkaline sulfite treatment discussed above.

Since water diffusion requires a lenghty treatment process, the water must be inhibited in order to prevent the metal from rusting. Alkaline chemicals, such as a 5 percent sodium sesquicarbonate, 5 percent sodium carbonate, or 2 percent sodium hydroxide solution will inhibit rust formation on the surface of the artifact; only the water, however, will remove the soluble chlorides. Chloride-contaminated iron should be immersed in successive baths of tap water until the chloride level of the bath approximates that of the tap water; the artifact should then be placed in successive baths of de-ionized water until the chloride count in the bath has leveled off and ceases to rise when it is changed. Uninhibited de-ionized water should not be used in water diffusion because it is very corrosive.

Organ (1955) suggests that alternating hot and cold temperatures will speed up chloride removal by alternately expanding and contracting the capillaries in the metal and the corrosion layer, causing a flushing action that will expel and draw in fresh water. From a standpoint of coefficients of thermal expansion, however, the alternate heating and cooling most likely changes the diffusion gradient of the solution rather than significantly changing the size of the metal capillaries. In some instances, this will decrease the time required to remove the soluble chlorides, but Oddy and Hughes (1970:187) found that there was no significant difference between the time required to wash similar objects of both iron and bronze at room temperature and at ca. 50°C. The alternate heating/cooling cycle may facilitate the removal of chlorides, however, when significant levels of chloride are present, and months to years are required to remove them. After the soluble chlorides have been eliminated, the artifacts must be carried through the same final steps as iron treated by other methods.

Following any conservation treatment -- electrochemical, electrolytic, chemical, or water diffusion -- it is necessary to remove insoluble oxide sludge, metallic powder, residual chlorides, and chemical residue through intensive rinsing (Plenderleith and Werner 1971:20). In electrolytic reduction or water diffusion, the artifact is removed after establishing that the chloride count in the solution has leveled off and ceases to rise when it is changed. The artifact is then rinsed thoroughly in several changes of alternate boiling and cold de-ionized water to remove any residual electrolyte and chlorides. Boiling water rinses oxidize the surface of the metal to a flat, pleasing black color. Iron artifacts may rust in de-ionized water during prolonged periods of rinsing; this can be prevented by adding gluconic acid, sodium gluconate, or sodium glucoheptanate to the rinse water. The gluconates act as rust inhibitors during the rinsing process and continue to serve in this capacity during solvent dehydration, heat drying, or air drying. Pearson (1972a:13-14) prevented Captain Cook's cannons from rusting during the rinse process by rinsing with a potassium chromate solution (1000 ppm chromate) with a pH not lower than 8.5. The strict disposal requirements for chromate solutions, however, prevent their large-scale use. Neither the gluconates nor chromate solutions are used widely, and satisfactory results are achieved without them. Nonetheless, conservators should be aware of the protection that they can provide, when and if their use is required.

The artifact should stand in the last vat of rinse water for a minimum of 24 hours. A sample of the bath water is taken and acidified with nitric acid and tested with 0.2 N silver nitrate for the presence of chloride. The silver nitrate test is suggested because it is quick, qualitative, and quite sensitive to minuscule amounts of chlorides. If the test is positive the artifact is returned either to electrolysis or further rinsing. If the test is negative, the artifact is ready to be dried and sealed with microcrystalline wax.

Specimens treated by water diffusion are put through a similar rinsing process. Since many of the objects treated by water diffusion are, however, very fragile, they may not be able to withstand the mechanical action of boiling water. The rinse water, if heated, should be kept below the boiling point for such artifacts, and sodium glucoheptanate should be added to the rinse water as a rust inhibitor.

The presence or absence of chlorides is determined by the silver nitrate test (Plenderleith and Werner 1971:201). The artifact is placed in distilled or de-ionized water for a few hours or overnight. A 10- to 20-ml sample of the solution is placed in a test tube and acidified with a few drops of dilute nitric acid (ca. 10 percent). The solution is mixed and five drops of 0.2 N silver nitrate (17 g of AgNO3/1 liter of H2O) is added. The test tube is held against a black background with good side lighting. If any chlorides are present, a white opalescence will be apparent. Under ideal conditions, with clean glassware and uncontaminated reagents, the test provides a good qualitative indicator for the presence or absence of chlorides.

After rinsing, the moisture absorbed by the artifact must be removed before any sealant, with the exception of certain waxes, which are heated above the boiling point of water, can be applied. When specimens can be immersed in any wax heated above the boiling point of water, drying is an optional step. Artifact drying can be accomplished by heat, vacuum desiccation, or dehydration in water-miscible alcohol or acetone. After treating iron, the metal surfaces are in a reactive state and quickly rust on exposure to air. In order to prevent the rapid formation of superficial rust, contact with air should be minimized until tannic acid or the final sealant or insulating coating is applied to the surface of the artifact. Some exposure to air is inevitable, and it is particularly troublesome when drying by heat (ovens or infra-red lamps) or vacuum desiccation. Infra-red lamps are not very effective on dense objects, and it is expensive to obtain ovens or vacuum chambers to accommodate very large specimens.

An alternative is to use a water-miscible solvent, ethanol, methanol, isopropanol, or acetone. Isopropanol is recommended because it is non-toxic, has a high flash point, and does not have an obnoxious odor. Ethanol and acetone are as effective, or even more effective, as isopropanol but are toxic and have lower flash points. Each of these solvents prevents the problems of rusting when exposed to air and can be used on objects of any size. Drying in an oxygen-free environment, such as that provided by alcohols, is ideal for objects with little metal remaining, since it will prevent the remaining metal from rusting and ferrous compounds from oxidizing to a ferric state. Both reactions will cause artifacts to expand and slough off the oxide layers. Alcohols also have the advantage of enhancing the removal of any remaining soluble chlorides and water in the specimens. In addition, all stains and undesirable features can be removed by brushing them while the objects are still in the alcohol. Artifacts also can be stored indefinitely in alcohol until it is convenient to process them.

When rinsing is complete, the artifact is removed while the water is hot and wiped with rags. This allows most of the water to evaporate. The artifact is immersed in alcohol that has been previously used for drying wet objects in order to remove the bulk of the remaining surface water. It is then submerged in water-free isopropanol to dehydrate for a minimum of 24 hours. By taking these precautions, the water content of the isopropanol bath will remain low and it can be thus used repeatedly. When the water content of the isopropanol becomes sufficiently high, it is used for the preliminary rinse, and fresh alcohol is used for the dehydration bath. This efficiency procedure is important during periods of material shortages and high prices.

It is imperative that the surfaces of treated marine-recovered iron artifacts be covered with a protective coating to insulate the metal from the effects of moisture, chemically active vapors, and gases. The right sealant or coating should be chosen to provide a protective moisture barrier and prevent corrosion. In general, the sealant selected should be (1) impervious to water vapors and gases, (2) natural-looking, so that it does not detract from the appearance of the artifact, (3) reversible, and (4) transparent or translucent, so any corrosion of the metal surface can be quickly detected.

Various monomers, acrylates, acetates, epoxies, paints, oils, lacquers, and other sealants have been used in the past, but few have withstood the test of time. Many of these sealants craze, peel, are irreversible, or have a high degree of permeability to water vapor. No single sealant is completely successful and all have some disadvantages, but microcrystalline waxes best satisfy the requirements of conservation. They have a high melting point and are relatively hard; in addition, they are the least permeable to water vapor of any of the sealants commonly used (Rudniewski and Tworek 1963:212). Besides sealing the surface of the artifact from the atmosphere and moisture, they provide considerable stability and strength to the object and are excellent for consolidating those which are fragile. In contrast, some laboratories (North 1987:230) use microcrystaline wax as the final coat only on cast-iron artifacts; its use on wrought iron that is to be stored and displayed indoors is also recommended.

Cosmoloid 80H is the most often-recommended microcrystalline wax mentioned in the conservation literature, but it is not available in the United States. Gulf 75 Micro-wax and Witco180M microcrystalline wax melt at approximately 180°F and make satisfactory substitutes. In some instances it may be advisable to dehydrate the artifact in alcohol before it is placed in a vat of microcrystalline wax; however, when microcrystalline wax is used as the final sealant, it is possible to eliminate the drying process for a great many iron artifacts. Artifacts can be taken directly out of the rinse water and placed in a vat of wax that is heated to 175°C, well above the boiling point of water. The artifact must be kept in the wax long enough and at a high enough temperature to completely vaporize the water and until bubbles stop evolving from the artifact. This may require several days for large artifacts. After complete penetration, the wax is cooled to 93-107°C, the artifact is removed, and the excess wax is promptly wiped off with rags. Since the water boils out of the artifact, this technique should not be used on fragile objects or objects with a loose oxide layer; fragile specimens should be dried by one of the methods mentioned above. With this exception, combining the water removal and the sealant steps remains a very satisfactory approach. Time and expense are saved, and good results are achieved.

The temperature at which the artifact is removed determines the thickness of the wax coating. Too low a temperature results in an obvious layer of wax, while at too high a temperature, all of the wax will run off the surface of the object. If any excess wax should remain after cooling, it can be removed with a torch, a hot air gun, or by scraping lightly with a knife. Scraping is the simplest method and leaves the least obvious scars on the wax film. Additional wax or wax with graphite added to it as a pigment can be used to cover surface defects in the metal and enhance the appearance of badly corroded objects.

In many laboratories, facilities are not available to impregnate large objects, such as cannons and anchors, with microcrystalline wax, so other coatings must be used. There have been experiments with chromate paints, lacquers, clear epoxies, linseed oil, and polyurethane. In general, all but polyurethane are ineffective. Over a period of months they craze, peel, and become permeable to moisture. The opaque coatings hide the surface of the artifact from view, preventing one from observing the corrosion occurring under the coating. In addition, the surface finish of the epoxy is too glossy and is irreversible, causing further damage to a few specimens which have to be re-treated.

Polyurethane-based paints or coatings are thermoplastic polymers that have many favorable attributes for serving as a protective coating on treated iron objects. They form clear, fast-drying, tough, flexible coatings with excellent adhesion and are highly resistant to moisture, salt water, acids, alkalis, abrasions, and weathering. The coatings can be removed with aromatic and chlorinated solvents, such as toluene or ethylene dichloride. Polyurethane comes in gloss and stain finishes. The gloss finish has more resin and is, therefore, more durable. It is recommended for outdoor use. The satin finish has less resin and has silica added to give a more acceptable flat finish, but it is less durable than the gloss finish and is generally recommended for interior use. By using an undercoat of gloss and a second coat of satin finish and by adding graphite to one or both coats, a very acceptable, translucent finish can be obtained that does not detract from the underlying surface color of the specimen being coated.

The use of polyurethane coatings in the manner described above or by themselves can be recommended for maximum protection of large iron artifacts to be displayed outside or in areas of high humidity and salt vapor in the air (North and Pearson 1975:177; Hamilton 1976:55; North 1987:230). For example, a large 18-lb. Civil War cast-iron siege cannon recovered from Galveston Island on the Texas Gulf coast needed to be sealed after treatment, but the cannon was larger than any of the wax vats in the laboratory. The surfaces of the cannon were painted with a 20 percent tannin solution to form a corrosion-resistant ferric tannate film. The painted surfaces were then left to air oxidize for two days. The cannon was then painted with a coat of clear gloss polyurethane, allowed to dry, and then painted with a coat of satin polyurethane. Graphite was added to both the gloss and the satin polyurethane to completely dull any glossiness to the surface finish. The results were very satisfactory.

Success has also been had with using Rustoleum, a fish-oil-based paint, but it has only a 10-year durability span, as opposed to 20 years for polyurethane (North 1987:230). For sealing wrought-iron artifacts that are to be displayed indoors, North (1987:231) recommends using clear- drying zinc phosphate-based anti-corrosion primer as the first coat, followed by up to six applications of high-durability, clear acrylic lacquer, and finished off with a final coat of Krylon Matte Spray Finish (North and Pearson 1975:177, North 1987:230). For large objects that are to be displayed outdoors, Townsend (1972a:253) suggests using a mixture of three parts zinc silicate powder to two parts water. This mixture forms a light beige paint that oxidizes to a light nautical gray. The coating, being anodic, provides cathodic protection of the iron object and is said to be highly resistant to salt spray, rain, sunshine, and temperature fluctuations. Large anchors and other implements painted with zinc silicate have been displayed outdoors in North Carolina without damage for many years.

With the exception of microcrystaline wax, which is easily removed by placing the artifact in a vat of boiling water, all other sealants present some problems in terms of reversibility. Polyurethane must be sandblasted off, and Rustoleum can only be removed with sodium hydroxide. For ease of application, resistance to water vapor, transparency, and the ability to strengthen the surface of the artifact, microcrystaline wax is the best sealant for both cast-iron and wrought-iron objects that are stored and displayed indoors. If an object is to be displayed outdoors, or if it cannot be treated with microcrystaline wax, polyurethane-based paints are recommended.

Artifacts that are so badly corroded that they cannot be treated, and compound objects with metal and organic parts requiring treatment but which cannot be separated, can be embedded in clear plastic blocks. Smith and Ellis (1961:32-35) describe the process of embedding a wrought-iron swivel gun and a Spanish battle sword in Selectron 5000 resins. This technique is drastic, with no hope of ever extracting the artifacts from the blocks, but it remains a possibility for very select, problem artifacts.  

The conservation of artifacts should produce objects that are chemically stable with an aesthetically acceptable appearance. Treatment should be reversible in the event that the object should require additional preservation. Re-treatment is generally prevented only if the artifact is stored or displayed under optimum conditions. Atmospheric pollutants, sulfur dioxide, hydrogen sulfide, sodium chloride, dust, and soot are detrimental, ubiquitous, and difficult to control even inside a reasonably sealed building. The relative humidity in which an artifact is stored is a particularly critical factor in its stability. The moisture level at which corrosion appreciably accelerates is called the critical humidity and is considered to be 60 percent for iron and steel (Cornet 1970:443). If iron still contains chlorides, a humidity as low as 50 percent may have to be maintained. Subsequent corrosion sometime in the future is inevitable with higher relative humidities. All the potential corrosive factors should be taken into consideration when storage facilities are being planned.

Since metal artifacts can eventually become chemically unstable from a myriad of causes and may need additional treatment, periodic inspections and evaluations of the artifacts are necessary. A conserved artifact of iron from a marine site remains a piece of metal, just as susceptible, and in fact more susceptible, to continued corrosion as any other piece of iron. Proper conservation does not ensure that an object will be preserved in perpetuity. At our present stage of knowledge, perhaps it is most realistic to say that the objective of artifact conservation is to delay eventual re-treatment for as long as possible and to make any necessary treatment simple and brief. There remains a lot of room for improvements in the conservation of iron. Still, at the present state, there are a number of procedures that successfully contend with the majority of problems encountered.

The majority of iron artifacts recovered from marine sites is treated by electrolytic reduction. This treatment consistently produces stable artifacts with a minimum of inexpensive equipment and chemicals, as well as a minimum of 'hands-on' treatment time. The alkaline sulfite treatment is commonly used for artifacts with badly corroded surfaces that could possibly flake off during electrolysis. Less common, but still used by a few laboratories, is hydrogen reduction of iron, but the cost of the equipment prohibits it more general use; this also applies to hydrogen plasma reduction techniques. Although not a consistently reliable treatment, various forms of intensive rinsing are sometimes employed on problem artifacts, and usually in conjunction with other treatments.

To carry out the above treatments, adequate space and equipment are required. Essential equipment includes various regulated DC power supplies, plastic and metal vats, anode material, wire, clips, fume hoods, pneumatic chisels, air compressors, a fork-lift truck if heavy artifacts are to be treated, a source for heating the rinses and wax, a pH meter, and an X-ray machine. All are easily secured at reasonable expense, and much of the equipment can be secured through federal surplus.

Last updated: