Furnace atmosphere

The recommended furnace atmosphere conditions necessary for good brazing are as follows:

• Dew point: ≤ -40°C
• Oxygen: < 100 ppm
• Inert gas: nitrogen

The most common source of nitrogen is that generated from liquid nitrogen storage tanks. A typical nitrogen gas specification from a liquid source indicates that the moisture content is <1.5 ppm (dew point = -73°C) and an oxygen level of <3 ppm. In brazing furnaces however, the normal atmospheric operating conditions almost always exceed incoming nitrogen contaminant levels. This is due to water and oxygen dragged into the furnace by the incoming products, by the stainless steel mesh belt and by the potential back-streaming of factory atmosphere through the entrance and exit of the furnace. The latter will occur when the exhaust and incoming nitrogen are not properly balanced.

Many furnaces are equipped with dew point and oxygen measurement devices. It is important that the measurements are taken in the critical brazing zone of the furnace because this is where these impurities will reach their lowest concentrations. Measuring dew point or oxygen levels anywhere else in the furnace may be of academic interest, but will not represent actual brazing conditions.

Dew Point Measurement

Measuring the moisture content in the critical brazing zone of the furnace has always been a key indicator of the quality of the brazing atmosphere. Moisture can substantially influence the quality and appearance of the brazed heat exchanger as well as the first time through braze quality (% rejects).

Chilled Mirror Technology

One of the more common principles of measuring dew point is using chilled mirror technology. The measurement of the water vapor content of a gas by the dew point technique involves chilling a surface, usually a metallic mirror, to the temperature at which water on the mirror surface is in equilibrium with the water vapor pressure in the gas sample above the surface. At this temperature, the mass of water on the surface is neither increasing (too cold a surface) nor decreasing (too warm a surface).

In the chilled-mirror technique, a mirror is constructed from a material with good thermal conductivity such as silver or copper, and properly plated with an inert metal such as iridium, rubidium, nickel, or gold to prevent tarnishing and oxidation. The mirror is chilled using a thermoelectric cooler until dew just begins to form. The temperature at which dew is formed on the mirror is displayed as the dew point.

The advantage of the chilled mirror dew point meter is that it is an absolute measurement with high precision. However, this measurement technique is sensitive to pollutants and corrosive contaminants which, in the brazing process, include KAlF4 condensation and trace amounts of HF gas. Consequently, the mirror requires frequent maintenance and replacement. “Dirty” mirrors can lead to false readings.

Coulometric Measurement Principle

The principle of operation for measuring is that an electrolyte is formed by absorption of water on a highly hygroscopic surface (e.g. P₂O₅) and the current level obtained to electrolyze the surface is proportional to the water content. The advantage of this principle of operation is that it is insensitive to aggressive media. The disadvantage is that the precision is not as high as chilled mirror technology. Some heat exchanger manufacturers have reported good success using this measurement principle in their CAB furnaces.

Relationship between dew point and moisture content

The relationship between dew point and moisture content is not linear. It is important to note that small changes in dew point will result in large changes in actual moisture content. This is evident from the graph shown below.

As manufactured, a non-corrosvie K-Al-F-type flux typically is a mixture of potassium tetra-fluoroaluminate (KAlF₄), and also contains potassium penta-fluoroaluminate (K₂AlF₅). K₂AlF₅ exists in different modifications: potassium penta-fluoroaluminate hydrate (K₂AlF₅ · H₂O), and hydrate-free (K₂AlF₅).
During the brazing process, the material undergoes essential physico-chemical alterations. While the chief component, KAlF₄, is simply heated up, the compound K₂AlF₅ · H₂O begins to lose its crystal water from 90°C (195°F) on. When the temperature is further increased within the ranges of 90° – 150°C (195°F – 302°F), and 290°C – 330°C (554°F – 626°F), two different crystallographic (structural) modifications of K₂AlF₅ are formatted.

When the furnace temperature is raised above 490°C (914°F), K₂AlF₅ begins to react chemically. According to the equation:

2 K₂AlF₅ → KAlF₄ + K₃AlF₆ (Equation 1)

the exact amount of potassium hexa-fluoroaluminate (K₃AlF₆) necessary for a eutectic flux composition (i.e. mixture of two or more substances which has the lowest melting point; see phase diagram) is obtained from the original K₂AlF₅ content. At brazing temperature, the resulting flux composition has a clearly defined melting range of 565°C to 572°C (1049°F – 1062°F). The flux melts to a colorless liquid.
Due to a vapor pressure of 0.06 mbar at 600°C, some of the KAlF₄ evaporates during the brazing cycle, particularly once melting temperature is reached. The total content of KAlF₄ contained in the exhaust is depending on time and temperature. Based on results from TGA analysis (with a heating rate of 20°C/min), the quantity of volatile compounds in Flux between 250°C and 550°C (482°F and 1022°F) is approximately 0.2 to 0.5%. These flux fumes contain fluorides and have the potential to react with the furnace atmosphere, especially moisture, to form hydrogen fluoride according to the equation:

3 KAlF₄ + 3 H₂O → K₃AlF₆ + Al₂O₃ + 6 HF (Equation 2)

This is one of the reasons, why the brazing process should take place in a controlled atmosphere (nitrogen) with low dew point and low oxygen level (another reason is to minimize re-oxidization effects on the aluminum surfaces).

Directly after brazing has been completed, flux residues consist mainly of KAlF₄ and K₃AlF₆. In the presence of moisture from the surrounding atmosphere, the K₃AlF₆ is converted back to K₂AlF₅ · H₂O over time (several days) in a reaction reverse to the one described in equation 1 followed by a re-hydration step.

The schematic below illustrates the transformations that occur as the flux is heated to brazing temperature. Note that these phases are unstable outside the furnace atmosphere.

European Association for Brazing and Soldering (EABS) and Solvay Fluor announce the tenth presentation of their joint technical training seminar entitled:

The theory and practice of the flame- and furnace brazing of aluminium

Dates:
5th and 6th October 2011

Venue:
Solvay Fluor GmbH
Hans-Bockler Allee 20
30173 Hanover
Germany

Purpose of the Seminar:
This technical training seminar will be presented at the Conference Centre and laboratories of Solvay GmbH, in Hannover, Germany. It will provide information concerning the manufacturing practices commonly used for brazing operations and, in particular, will address the three fundamental aspects of the industrial-scale brazing of aluminium.
These are:

• The flame brazing of aluminium.
• The Controlled Atmosphere Brazing (CAB) of aluminium heat exchangers with non-corrosive fluxes (NOCOLOK^® Flux).
• The methodology of how to ensure that the brazing process selected is, indeed, the one that represents best practice.

Who should attend this two-day seminar?

• Technical staff who need to have a specific understanding of either one or both of the fine details of the technology of the brazing of aluminium with flames, and/or the NOCOLOK^® furnace brazing process.
• Design and production engineers who are fabricating, or who intend to fabricate, aluminium pipe-work assemblies and/or condensers and/or evaporators.
• Production Engineering Department Managers whose duties include day-to-day responsibility for the brazing of aluminium.

More Information, Programme and Booking:
EABS Secretariat
5 Kent Drive, Congleton. Cheshire, CW12 150, England
Telephone: +44 (0) 1260 2717 03
Telefax: +44 (0) 1260 27 67 29
e-Mail: eabs@btconnect.com
www.brazingandsoldering.org

Very often, heat exchanger manufacturers increase the flux loading on components to be brazed to compensate for furnace atmosphere or other process related deficiencies. The flux is an excellent “band-aid” and can be used as such, but only while the true problems are located and rectified. Long term use of higher than recommended flux loads can lead to other problems.

Over fluxing causes more KAlF₄ evaporation and condensation. This will load up the dry scrubber more quickly. White powder will accumulate more quickly on the curtains at the exit end of the furnace. If this is noticed, there is a very good chance that the dry scrubber is loading up more quickly.

There will be a more rapid build-up of the flux inside the furnace. This is a common issue with over fluxing whereby flux builds up on the muffle floor at the entrance to the cooling zone where it will solidify. This flux build up has been known to deflect the mesh belt.

There is more rapid build up of the flux on the fixtures which can significantly reduce maintenance intervals.

Over-fluxing can lead to visible flux residue on the brazed heat exchanger which may increase the incidence of flux residue fall-off. Excess flux residue dulls the appearance of a brazed heat exchanger and can also accumulate in the gasket areas causing problems with seals. Too much flux residue will also inhibit surface treatments such as painting or conversion treatments.

The theoretical amount of flux required to dissolve a 100 Å oxide film is about 0.02 g/m²
(1 Å = 10^-10 m = 0,1 nm). For a 400 Å film, still only 0.08 g/m² flux is required. These do not take into account losses to moisture, oxygen or poisoning of the flux by Mg alloy additions.

In practice however, the recommended loading for fluxing is 5 g/m², uniformly distributed on all active brazing surfaces. This is more than 250 times the theoretical amount required for oxide dissolution. To visualize what 5 g/m² flux loading might look like, think of a very dusty car. As the heat exchange manufacturer gains experience with his products, he may find that a little more is required for consistent brazing or that he can get away with a little less flux.

Too little flux will result in poor filler metal flow, poor joint formation, higher reject rates, and inconsistent brazing. In other words, the process becomes very sensitive.

Too much flux will not affect the brazing results. However there will be pooling of flux which can drip on the muffle floor, the surface of the brazed product will be gray and there will be visible signs of flux residue. Furthermore, flux will accumulate on fixtures more rapidly which then requires more frequent maintenance. More importantly yet, using too much flux will increase the process costs.

In some cases, heat exchanger manufacturers use higher than recommended flux loadings to mask furnace atmosphere deficiencies. This should be viewed as a short-term solution and the furnace problems should be addressed.

See also: How to evaluate flux load?

Dust and dirt, condensates, lubricants and oils must be thoroughly removed. If the metal work pieces are poorly prepared, the flux will not spread evenly and the flow of filler alloy will be haphazard: it will either not spread properly or will discolour. The consequence would be an incomplete joint.

The first step is therefore: always clean the components of all oil and grease. The surfaces can be cleaned using either chemical, water-based or thermal cleaning techniques and substances.

Aqueous Cleaning

Aqueous or water based cleaning is a quite efficient and robust process, but still generates some waste water.

Aqueous cleaning starts off with a concentrated metal cleaning agent, which is subsequently diluted with water to 1% to 5% (v/v). The composition of a supplier’s cleaning solution is proprietary, but usually contains a mixture of surfactants, detergents and active ingredients such as sodium carbonate that serves to elevate the pH. Once diluted, the cleaning solution will typically have an elevated pH in the range of pH 9 to 12. There are acid based solutions, but appear to be less common.

The best water-based cleaners contain water, tensides, cleaning agent and active ingredients such as carbonates.

The cleaning solution works best at higher temperatures and is usually recommended to operate at 50°C to 80°C. Cleaning action is quicker at higher solution temperatures.

Thermal Degreasing

Thermal degreasing works by elevating the temperature of the work piece so that lubricants present on the surfaces will be evaporated. This procedure only works with special types of lubricants known as evaporative or vanishing oils. Vanishing oils are light duty lubricants used mostly for the fabrication of heat exchanger fins, although they are now finding uses in the stamping and forming of other heat exchanger components. Lubricants not designed for thermal degreasing must not be used. These could leave behind thermal decomposition products and carbonaceous residues which at higher level prevent brazing and have the potential to degrade product appearance and accelerate corrosion.

Brazing sheet comprises of a core alloy clad on 1 or 2 sides with a lower melting aluminum-silicon (Al-Si) alloy. This thin layer, usually makes up 5 % to 10 % of the total thickness of the brazing sheet.

It melts and flows during the brazing process, to provide upon cooling a metallic bond between the components. It is common that the braze clad alloy are from the AA 4xxx series or more particularly AA 4343 (Al-6.8~8.2 wt.% Si).

However, if larger fillets are desirable, or if in a situation where brazing is likely to occur at lower temperatures, AA‑4045 is the preferred choice.

Manufacturing

Cross-section morphology of a double-side clad tubestock.

Aluminium brazing sheet is manufactured by roll-bonding techniques to clad a core alloy ingot on one side or both sides with a low melting AlSi alloy. As an alternative, one side can be clad with a non-braze alloy, e.g. Zn-containing alloy.

Depending on the desired final properties, the core is either homogenized or not before the cladding operation.
The whole package is subjected to preheating, hot rolling and cold rolling down to the final thickness of the respective products. Depending on the requested final properties, the material is subjected to final annealing and / or intermediate annealing operation(s).

The core provides structural integrity. It is common to use a variety of aluminium alloys, examples being AA 3xxx (AlMn), more particularly AA 3003, AA 3005, AA 3105 and modified versions of AA 3003 or AA3005 (long life alloys).

Phase diagram

Aluminum End of Al-Si Phase Diagram

The melting characteristics of the cladding alloys are governed by the Al-Si phase diagram. The eutectic composition, i.e. the amount of Si required to produce the lowest melting point is 12.6%. The melting point at this composition is 577°C. At lower Si levels the solidus or the point at which melting begins is also 577°C. However, melting occurs in a range and the temperature above which the filler is completely molten is called the liquidus. In between the solidus and liquidus, the filler is partially molten, existing both as liquid and solid. The difference between the solidus and the liquidus forms the basis for various filler metal alloys.

The table below shows the solidus and liquidus of common brazing alloys.

The higher Si alloys (e.g. AA4047) have higher fluidity and a narrower melting point range while the lower Si alloys have less fluidity with a wider, higher melting point range. Erosion of the base metal occurs when the braze alloy dissolves part of the core alloy. The extent of erosion is increased by:

• Higher Si levels in the braze alloy
• Longer braze cycles
• Excessive peak brazing temperatures
• Excessive thickness fo the braze metal layer
• A design which allows pooling of the braze metal to occur

For added strength and machineability, certain alloys contain Mg. Most notably are the 6XXX series alloys (up to 1% Mg) that are used for fittings and machined components and the so-called long life brazing sheet alloys (up to 0.3% Mg in the core). There is a limit to the amount of Mg tolerated in NOCOLOK^® Flux brazing. Up to 0.5% Mg can be tolerated in furnace brazing while around 1% Mg is tolerable for flame brazing.
When an Al alloy containing Mg is heated, the Mg diffuses to the surface and reacts with the surface oxide to form MgO and spinels of MgO:Al₂O₃. The diffusion is time-temperature dependent and is rapid above 425°C. These spinel oxides have reduced solubility in the molten flux. Furthermore, Mg and/or MgO can react with the flux forming compounds such as MgF₂, KMgF₃ and K₂MgF₄. All of these serve to poison the flux and significantly reduce its effectiveness.
In flame brazing, higher Mg concentrations can be tolerated since the faster heating rates do not allow the diffusing Mg enough time to appreciably decrease the beneficial effects of the flux. Flame brazing components containing > 1% Mg may be possible under some circumstances with increased flux loadings and very fast heating rates (<20 second braze cycle).
It should be noted that when one speaks of the brazing tolerance to Mg, it is the total sum of the Mg concentrations in both components:

[Mg] _{component 1} + [Mg] _{component 2} = [Mg] _total

The figure below shows the effect of Mg on fillet size and geometry:

0.1% Mg

0.4% Mg

If the user is experiencing difficulties brazing and suspects elevated Mg levels as the cause, there are a couple of ways to be sure. First, check with the supplier of the alloys or perform a chemical composition analysis on the suspect alloys. This is the most certain way. Secondly, look for a golden hue on the brazed product. This is an indication that Mg alloys are being used and the color is a result of the increased oxide thickness. Furthermore, there may be a very light, almost fluffy residue on the brazed component that can literally be blown off by mouth. These visual indicators can most certainly be traced back to poor brazing results due to the presence of Mg.

Improving brazeability
There are a few ways in which the brazeability of Mg containing alloys can be improved:

1. Increasing the flux loading. A substantial improvement is gained when increasing the flux loading up to 10 g/m² or more in furnace brazing. In cases where there is just one component containing Mg such as in a fitting, extra flux can be brushed around the area of the joint.
2. Increasing the heating rate. Slow heating rates allow more Mg to diffuse to the surface thereby hindering brazeability. For furnace brazing Mg containing alloys, the fastest possible heating rates achievable without sacrificing temperature uniformity will increase the tolerance to Mg.
3. Combining increased flux loadings and faster heating rates.
4. Maintaining proper gap tolerances and joint designs.
5. Increasing the nitrogen flow rate to minimize furnace atmosphere contaminants that also compete to reduce brazeability.

Tip: NOCOLOK^® Cs Flux

Better results are reported when using cesium containing fluxes for aluminum alloys containing Mg up to 0.6 – 0.8% Mg. Fewer leaks are observed when compared with standard flux and less porosity is noted in the joint areas. Furthermore, standard flux loads and braze cycles can be used with Cs containing fluxes.

NOCOLOK^® Cs Flux is a flux of the general formula K_xCs_yAlF_z where Cs is chemically bound. It has a melting range of 558°C – 566°C. The maximum Cs content is limited to 2% to keep the cost of the flux down. Increasing the Cs content does not increase brazeability as shown below:

Cesium reacts as a chemical buffer for Mg by forming CsMgF₃ and/or Cs₄Mg₃F₁₀. The flux inhibiting factors of Mg are therefore reduced.

How to measure?

In the case of heat exchangers, the surface area being fluxed must first be determined. For ease of calculation, the louvers on the fin can be ignored. The radius on the fin can also be ignored.

Imagine then the fin pulled out of the heat exchanger and straightened out to form one long strip. Similarly, the surface area of the slots in the header can also be ignored.

Remember that in calculating the surface area of the heat exchanger, there are 2 sides to every tube, 2 sides to every fin and 2 sides to the headers. The total surface area is then expressed in m2: All dimensions are in meters (m) to yield a surface area in square meters.

Header

Assuming it is a cylindrical (condenser) header:

SA (m2) = (2 x 3.14 x radius of header(m)) x length of header (m) x 2 headers

Assuming it is a radiator header:

SA (m2) = length of header (m) x width of header (m) x 2 (sides/header) x 2 (headers)

Tubes

SA(m2) = width of tube(m) x length of tube (m) x 2 (sides/tube) x total number of tubes

Fins

Ignore the louvers in the fins

SA (m2) = width of fin (m) x (fin height (m) x number of fin legs/tube) x 2 ( sides/fin) x total number of fins

Total Surface area in m2 = SA headers + SA tubes + SA fins

To determine the flux loading, a degreased and thoroughly dry heat exchanger is weighed. The heat exchanger is then run through the fluxer, blow-off and dry-off section of the furnace. The heat exchanger is removed just prior to entering the brazing furnace and weighed again.

The flux coating weight is then determined using the following formula:

Weight of unit fluxed and dried (g) – weight of unit un-fluxed (g) x Surface area (m2)

To make sure that the flux loading was determined on a completely dry unit, run it through the dry-off section a second time and re-weigh.

See also: Flux loading