For our products we are looking to use laminates that have the correct electrical, mechanical and thermal properties, but also sufficient chemical resistance (primarily to withstand the fabrication and assembly processes), humidity resistance (for an extended life) and fire retardance (to meet commercial and legal requirements).
These parameters are affected by the manufacturing method, structure and materials. For example:
A specification sheet for a laminate is an extended document! Good examples of the genre are the BS EN 60249 specifications for generic FR-2 and FR-4 materials. These specify:
1 Surface Insulation Resistance is a measure of the unwanted conductivity across the surface. How this is often measured is described in the context of flux in How joints are made.
Non-electrical properties of the clad sheet
Non-electrical properties of the laminate after removal of the copper
There are also requirements for packaging and marking, and recommendations for choosing which acceptance tests will be carried out by the purchaser of the material.
At this point, if you have time, we would recommend you to take a look at typical specifications, and see the extent to which they contain comparable information. One suitable site which lists a range of laminate materials is http://www.isola.co.uk/: look under ‘Products’.
There are minor variations between test specifications from different institutions, but a very convenient source, and one we often make reference to, is the IPC-TM-650 Test Methods Manual. All the tests in that manual can be downloaded without charge (though as separate .PDF files) from http://www.ipc.org/.
In this document, we are reviewing these properties in generic terms, looking in detail at design, test and application aspects, but specifically focussing on FR-2 and FR-4 laminates. [A wider range of materials is discussed in Alternative board materials]
[ back to top ]
Figure 1 is an attempt to pull together the important physical and mechanical issues for a board design, looking at all aspects other than the thermal and electrical characteristics, which are considered separately.
Table 1 indicates typical maximum temperatures quoted. Note that, as in this case for FR-4, a distinction may be drawn between the maximum operating temperature for adequate electrical function and a slightly higher temperature for acceptable mechanical performance.
|operating temperature °C|
|material||electrical parameters||mechanical parameters|
Use a web search to try and verify the operating temperature range for FR-4 laminate.
Search hint: you will get more (and more useful) results if you search for "characteristics" rather than "operating temperature range". This is an example of where a web searcher needs to think laterally!
Then try a second search using "epoxy-glass laminate" instead of FR-4. How do the results compare? Can you explain any differences?
[Note, when you are searching, that FR-4 and FR-4 will yield approximately equivalent numbers of hits, most of which are not duplicated! Also be careful that you use " either side of the string as the delimiter, rather than “ ”.]
As the temperature of a laminate passes through its glass transition temperature (Tg, sometimes referred to as ‘T sub g’), molecular bonds begin to weaken, and the resin changes from a ‘glassy’ state to a more random ‘rubbery’ state. The transition is fully reversible– as with water turning to ice and back again – and the only permanent change that will take place is if the ‘rubbery’ material is allowed to distort, and then ‘sets’ in this position when cooled.
The ‘phase transition’ phenomenon, which is seen in a wide range of materials, both thermoset and thermoplastic, is accompanied by changes in both dimensional stability and a range of properties. Exceeding Tg is typically accompanied by substantial increases in CTE, reduction in flexural strength and changes in electrical properties, and for this reason Tg is often regarded as being synonymous with maximum operating temperature.
Tg is a measure of how well a laminate resin system resists softening from heat, but is just the temperature at which molecular bonds begin to weaken enough to cause a change in physical properties: this is not the same as saying that the resin reaches any kind of melting point during its glass transition. After all, resins retain some stiffness even above their Tg, whereas a molten material will not resist deformation to any significant extent.
To understand the difference between glass transition and melting, we need to recall that some polymers are crystalline and others classified as ‘amorphous’. If you find difficulties here, then look up the topic in Polymer basics.
Crystalline polymers melt when the polymer chains are released from their crystal structures and become a disordered liquid. On the other hand, amorphous polymers do not have a single temperature at which the crystalline structure ‘collapses’. Instead, the polymer chains can move more easily as the temperature increases, but they are never released totally from their micro-molecular structure.
Whilst one might expect a clear distinction between the two types of polymer, the position is complicated by the fact that even crystalline polymers have some amorphous proportion which might make up 40–70% of the polymer sample. Whilst the crystalline portion can undergo melting, the amorphous portion can only undergo glass transition, which is why many real polymers have both a glass transition temperature and a melting temperature!
Confused? Just remember that almost all materials currently used are thermoset materials, and as such have amorphous structures, so that the change in characteristics that takes place is gradual, over a range of temperatures.
The original way of determining glass transition temperature makes use of the fact that any phase change is accompanied by either generation or absorption of heat. [This is the same phenomenon which gave rise to the plateau in the cooling curves used to generate the phase diagram of solder, and which we shall meet again in condensation soldering] The method of differential scanning calorimetry (DSC) is well-loved of polymer scientists and is carried out in an apparatus shown schematically in Figure 2.
A sample of polymer, and an identical but empty reference vessel, are both heated to achieve a constant rate of rise of temperature. If the polymer in the sample pan neither generates nor absorbs heat, then the heaters under both sample pan and reference pan will require the same power in order to heat them to the same temperature. However, if the polymer sample either gives out or absorbs heat, the amount of heat to be supplied by the feedback control system, consisting of thermocouple controller and heaters, will automatically regulate the heat flow at a different level.
The equipment plots the difference in heat output for the two heaters against the temperature achieved. If the material is just heating up, then the rate of heat flow will be constant, and determined by the specific heat of the sample. When there are changes in state, as for example when the polymer crystallises or reaches its glass transition temperature, then there is either a dip or an increase in the plot. A permanent change in increase represents a permanent change in state. To learn more about this, why not take a look at the website run by the Department of Polymer Science at the University of Southern Mississippi at http://www.psrc.usm.edu/macrog/dsc.htm, where there is a nice explanation of some of the other things that DSC can do.
DSC is an indirect method of measuring glass transition temperature; thermal mechanical analysis (TMA) is a more direct method of determining the temperature at which the glass transition of a material makes a difference to the rate at which its dimensions change. The method is particularly appropriate for the dielectric materials used in printed boards, because it is easy to apply to changes in laminate thickness, and it is this Z-axis expansion that is particularly important in determining the reliability of a board structure.
The sample, with a quartz rod resting on its surface, is placed in a small heating chamber. As the chamber is heated, the material begins to expand, moving the quartz rod along with it. At the other end of the quartz rod is an extremely sensitive gauge that can measure movement, so that any change in dimension can be plotted against temperature. Quartz is chosen because it has a low CTE, and transfers the majority of the expansion from sample to gauge.
The actual measurement of the movement may employ different types of transducer, depending on the style of equipment:
Another way of observing mechanical changes in a material is to look at the way in which it transmits force. The so-called dynamic mechanical analysis (DMA) technique clamps the sample between a force generator and detector which is placed inside a heated chamber. As the material goes through glass transition, its ability to transmit force changes.
IPC-TM-650 Method 188.8.131.52 uses TMA to measure the dimensional change with temperature. Around the glass transition temperature there is a change in slope of the dimension-temperature curve, and the value of Tg can be estimated by the intersection of the tangents to the two slopes (Figure 3).
Note that the IPC test method specifies that:
In other words, be aware that measurements must be applied to the laminate itself, and not be overly affected by the presence of the copper foils, which have a different rate of expansion, and no discontinuity in CTE.
As explained in Coefficient of Thermal Expansion, materials change dimensions in response to changes in temperature, and we have just seen that polymeric materials which exhibit glass transition have different values of CTE depending on whether the measurement is taken above or below Tg. For reinforced materials, there will also be differences along the different axes, as a result of the reinforcement being in the plane of the laminate (the X/Y plane).
Typically, laminates have similar CTE in X and Y directions, corresponding to the warp and weft directions of the woven material. However, this results from conventional weaves having broadly similar constructions in both directions, and is not necessarily the case. However, in the Z axis (through the thickness of the material) there is no constraint on TCE as a result of the reinforcement. As a result, the Z-axis CTE may increase substantially above Tg, and the expansion and contraction that occur with changes in temperature can lead to deformation in the through-holes and stress on the surface pads. Over time, thermal cycling can fatigue the through-hole and cause failure by separation of the conductor from the hole wall or cracking of the conductor within the barrel. This is the reason why attention is paid to creating a plated layer of sufficient ductility.
The CTE is a function of the materials used and the volume percentage of these materials. The key factors, particularly in Z-axis expansion, are the performance of the resin system, and the resin content of the laminate, as resin systems have relatively high CTE compared to reinforcements. Using more fillers will depress CTE; ‘resin-rich’ systems will tend towards higher values.
This Z-axis expansion has an impact on through-hole reliability (see Board failures), but be aware also that expansion in the plane of the laminate also becomes important when large components are attached to the surface: differences in CTE between board and component can compromise the reliability of the bond.
One possibility we will explore in Unit 10 is to control CTE in the X/Y plane in order to improve the reliability of certain structures. One way of lowering the CTE, and making it more compatible with silicon and ceramic materials, is to change the foil from copper (CTE ~17 ppm/°C) to a material of lower CTE. The most common replacement foil is a sandwich construction with copper outside and invar inside. Invar is a 36% nickel-iron alloy whose CTE is less than 1.3x10–6/°C over the range 20–100°C, and very low over the range –100°C to +200°C. Invar is commonly used in applications such as CRT shadow masks, and was the forerunner of a family of controlled expansion nickel-iron alloys used in bimetallic strips and thermostats. For more details, search for "invar" at http://www.azom.com. Having a copper surface makes the material compatible with normal PCB processing – the thickness of the skin on standard CIC foils is either 12.5% or 20% of the overall thickness.
Although the value of the glass transition temperature is a measure of the stability of a material when heated, it is not necessarily a guarantee of better performance above Tg. In theory at least, the properties of a high-Tg resin may change much more rapidly above Tg than one with a low Tg, and the lower Tg material may even exhibit superior properties at high temperature (Figure 4).
In practice, materials with a higher Tg are generally favoured for laminates because they maintain their dimensional stability over a wider temperature range, and the total expansion due to temperature up to solder reflow temperature is lower for a high Tg material (Figure 5).
This reduced total expansion has a significant benefit in reducing the strains associated with the copper barrel of a through-hole or via. Reduced strain results in reduced pad lifting, reducing copper barrel cracking, and reduced inner-layer foil cracking.
The increased Tg has other benefits:
For most laminates, the glass transition temperature is influenced mainly by the chemical structure of the resin: difunctional epoxy exhibits a Tg of 125–135°C; polyimide 250–260°C. With epoxies, there is a substantial spread in Tg between different materials (Figure 6), the value depending on the degree of cross-linking, so that it can be increased by using a multifunctional chemistry. However, the glass transition temperature of phenolic resins is related to the cure temperature, and can be elevated by baking.
We have concentrated primarily on the glass transition temperature, because this is so often used almost as a shorthand way of indicating the quality of the material. Of course this simplification has limitations, and we will find that other materials, although they have a high glass transition temperature, are not so satisfactory as epoxies from the point of view of fabrication.
What happens to a laminate at high temperatures will be considered in the next part of this unit. However, from the perspective of environmental resistance, we also need to consider what happens to a laminate material when it is exposed to:
Thermal shock and thermal cycling are both related to temperature change, the term ‘shock’ being used for rapid changes and ‘thermal cycling’ for the slower, more gradual changes that are typical of life. For the purposes of testing2, thermal cycling generally takes place in an air environment, with rates of change restricted to 5°C/minute, whereas thermal shock uses a rapid change of environment, and is often associated with liquid immersion in order to speed the transition.
2 The specification for change of temperature testing, BS EN 60068-2-14, describes three tests of increasing severity: Test Nb has a specified rate of change of 1, 3 or 5°C/minute; Test Na moves the part from one environment to another in 2–3 minutes; Test Nc involves liquid immersion. Advice for designers as to the applicability of particular severities of test is found in BS EN 60068-2-33. Note that care has to be taken that condensed moisture should not affect the results or impart additional stresses to the unit under test.
Over the normal operating range, most laminate materials will survive thermal excursions with little damage. However, when they are built into assemblies, associated solder joints may fail, particularly if changes in temperature are associated with warping of the board. As we will see later, this warping is generally caused by a lack of balance in the structure of the board, and as such is greatly influenced by design.
The resistance of a board to humidity and moisture is dictated primarily by the resin content, although moisture will be transmitted along the fibres of a reinforcement, particularly if the reinforcement is not completely wetted by the resin. This is one reason why great care is taken over surface preparation of glass fibres.
Laminates will typically shed gross droplets of liquid, but will absorb some moisture into their structure when exposed either to liquid or water vapour in the atmosphere. Typical polymeric materials will absorb around 1% by weight of moisture, a figure which can be verified easily by weighing before and after exposure. Whether or not this has an adverse impact on the reliability of the structure will depend on the ionic content of the resin – as we will see in later units, ions plus moisture plus volts can lead to electromigration. Some resins will also expand when they have absorbed moisture, potentially affecting the integrity of the board, but this is not usually a problem.
Finally, the ability of the board to withstand vibration and shock is mostly a design issue that is affected by the weight and distribution of the added components and the support provided by the enclosure.
Your new design has to use a lead-free solder with a melting point of 217°C, replacing eutectic tin-lead solder. Explain to your manager, who knows nothing about polymers, the reasons why you may need to use a more expensive laminate material for this application, and exactly what the differences between materials will be.
[ back to top ]
In Basic board materials we discussed machinability from the point of view of punching as against drilling, and seen how paper-based materials are more suitable for punching than are those containing woven strands of glass. Ease of punching can be improved by carrying out the operation on a warmed laminate, but care still has to be taken to specify a punchable grade where this is the intended method of profiling or creating holes.
The presence of woven strands of glass can also impact on the drilling process, both deflecting small diameter drills (reducing the allowable stack height) and reducing bit life through abrasion. The optical characteristics of the glass fibre also make it difficult to adopt the alternative approach of laser-drilling vias. Nevertheless, woven glass continues to be the most commonly specified reinforcement, because it combines reasonable ease of processing with adequate robustness and low cost.
An important characteristic for metals is the stress/strain ratio referred to as the modulus of elasticity or Young’s Modulus. [If you have forgotten what this means, look at Mechanical properties of metals]. Laminates too will have a Young’s Modulus, but are less likely than metals to be exposed to direct compression or tension. However, laminates will get flexed, so the most appropriate measure of the robustness of a material is its flexural strength.
The generic ASTM flexural strength test shown in Figure 7 measures the load which a beam of material will withstand without fracture when supported at the ends and loaded at the centre. The test piece is under compressive stress at its top surface and tensile stress at its under-surface, with a neutral axis along the mid-section of the beam.
For materials that do not break, the flexural strength is usually reported as the load at which 5% deformation/strain of the outer surface occurs.
If you bend a piece of laminate, you will find that it springs back exactly to its original state only for relatively small deflections. When greater loads are applied, and then removed, some permanent bending in the laminate will be observed. At higher strains still, the stress/strain curve may also show a reduced slope, similar to yielding in a metal, corresponding to fracture and delamination effects on a microscopic scale within the bulk of the laminate.
The test for the flexural strength of boards is defined by IPC-TM-650, Test 2.4.4, which is based on the generic ASTM-D-790 test. The standard specifies the nominal thickness and dimensions of the specimen, the span between supports, and the rate at which load is applied, and provides a formula for deriving the flexural strength. Be aware, however, that the IPC test is to destructive fracture, and provides no indication of the level of stress that leads to a permanent deformation of the laminate.
Typical values for flexural strength are given in Table 2. You will notice a slight difference between the lengthwise and crosswise values, which is another example of anisotropy. Reinforced composites of any kind will have maximum strength and stiffness in the direction of the alignment of the reinforcement. When half the strands are laid at right angles to the other half, the resulting strengths in either direction are less than those of the parallel alignment. Fortunately, most laminates are made with ‘balanced’ fabrics, with equal yarn distribution in warp and weft directions. The differences reported will probably be due to minor differences in the weave.
|FR-2||83 MPa||72 MPa|
|FR-4||414 MPa||345 MPa|
Both flexural strength and flexural modulus will be functions of temperature, with their values reducing with increased temperature, even below glass transition. It is important therefore that the board does not deform either under operation at elevated temperatures or during the more severe conditions of solder assembly, and this is why one IPC test (IPC-TM-650, Test 184.108.40.206) measures flexural strength at test temperatures that are comparatively close to the maximum working temperature, as shown in the following table
|resin type||thickness||test temperature|
|difunctional, tetrafunctional epoxies||all||125°C|
|hot strength retention epoxies high temperature epoxies||up to 1.6 mm 1.6 mm and over||190°C 170°C|
|BT-epoxies; modified epoxies||all||170°C|
|polyimide (Tg >250°C)||all||204°C|
|polyimides (Tg <=250°C)||all||170°C|
Board thicknesses are normally in the range 0.8–1.6mm, over three-quarters of products being in that range. However, there is a trend towards thinner boards. As boards become thinner, so the inherent stiffness of the laminate becomes more important.
Thin, flexible laminates are not easy to handle during fabrication. At assembly, there are particular difficulties at component placement, reflow and wave soldering, and for depanelling, as well as continuing difficulties when the assembly is mounted into its enclosure. Certainly the board may have to be supported in order to stay flat during both wave and reflow soldering:
These are issues we shall meet again when we consider soldering processes and the reliability of chip ceramic capacitors.
The three-point bending method used for flexural strength tests can also be used to determine the flexural modulus (resistance to bending) of the beam (Ef), from the ratio of the stress difference to the corresponding strain difference at two discrete points, normally corresponding to strains of 0.5% and 2%.
Referring to Figure 8, the flexural stress σf (in MPa) is calculated from the equation:
F = load applied at centre of beam (N)
L = distance between support points (mm)
b = width of beam (mm)
h = thickness of beam (mm)
The flexural strain εf is derived from the equation:
s = deflection at centre of beam (mm)
L = distance between support points (mm)
h = thickness of beam (mm)
The flexural modulus can then be calculated from the deflections s1 and s2 that correspond to two different values of stress, using the equation:
sf1 = flexural stress measured at deflection s1
sf2 = flexural stress measured at deflection s2
Note that these equations hold exactly only for linear stress/strain behaviour. For most plastics, they are accurate only for small deflections, which is why the two strains suggested are 0.5% and 2%.
Intuition, supported by calculation, will show that the flexural modulus for a board depends on the properties both of the laminate and the foil. The detailed calculations are complex, but an approximation for many mechanical properties of composites can be calculated using the ‘Rule of Mixtures’. Here the property of the composite is estimated as resulting from the responses of each component, weighted by the volume fraction of the component in the composite. For example, this gives a reasonable estimate of properties parallel to the fibres in a material. For more details see http://islnotes.cps.msu.edu/trp/back/est_rule.html and following pages.
The practical implication is that getting a stiff, thin laminate is not easy, unless we can use exotic materials, and our design has to allow for some flexing of the substrate.
As the discussion above has indicated, a key concept to bear in mind in relation to board flatness is that the laminate is not a stable material! The base laminate as supplied may have a fairly tight flatness tolerance, but by the time you have built up a number of layers made of materials with differing TCEs, and which flow and shrink during heat treatment, there are many opportunities for non-flatness.
The specifications distinguish between two types of deviation from flatness that are illustrated in Figure 9:
Bow, which is characterised by a cylindrical curvature of the board, so that all four corners are in the same plane, but the centre is raised.
Twist, which is deformation parallel to a diagonal, so that only three of the four corners of a rectangular sheet lie in the same plane.
Not only does excess distortion affect the fit of the final board in its enclosure, but warped boards create substantial difficulties for the assembler.
Basing your explanation on the previous discussion about flexural modulus, describe why it is that fixing an assembly with a warped board into an enclosure designed to take flat assemblies creates stress within the structure.
Would this be easier to do if the board were warmed first?
[ back to top ]
The important electrical properties of a PCB track (or ‘trace’) are summarised in Figure 10. In this section we shall be considering first the DC properties of an interconnect and then its reactive components, which are important at high frequencies.
The resistance of tracks varies in proportion to their length and inversely to their cross-sectional area. The track resistance R is given by the formula:
ρ = the volume resistivity of the conductor (approximately 1.72 × 10−6 W·cm)
L = the length of the conductor
A = the cross-sectional area of the conductor
This formula gives a reasonable estimate for the resistance of a wide track, but few narrow tracks have vertical sides. A better estimate of the shape is given by Figure 11, which shows a trapezoidal cross-section, where the undercut on each side is approximately half the thickness of the track.
Using this approximation, the area of the cross-section of the track can be estimated using the formula:
The thickness of the copper of course is the actual thickness of foil, together with any plating, allowing for the thickness tolerance on both plating and initial foil. Copper also has a substantial positive temperature co-efficient of resistance which may need to be taken into consideration if there is substantial track heating. For more information on calculating DC line resistance, see http://www.merix.com/resourcecntr/tech/dc_line.doc
Whilst for digital signals even the thinnest tracks that can be manufactured consistently and economically are more than adequate to carry the signal current, on high dissipation boards current-carrying capacity has to be considered. This is especially important with multilayer boards, which have higher track concentrations, and where tracks on inner layers cannot radiate heat direct to the air.
The current-carrying capacity of a track depends on its width and the thickness of the copper foil from which it was etched, and on the permissible heat rise, which is a complex function of materials, construction, operating temperature range and intended use. Figure 12 is a typical set of curves that shows how the steady state temperature rise in a 70 µm thick copper foil is related to current and track width.
IPC-2221 contains a fuller set of such curves, which may be used to calculate the recommended track width for any current. Obviously, in assessing what track temperature rise is allowable, the designer has to take into account the maximum operating temperature of the laminate and the worst case internal temperature of the equipment. The ways in which the full chart can be used, and a copy of the charts themselves is available at the UltraCAD website at http://www.ultracad.com/using_ipc_temp_charts.pdf
The charts rely on graphical interpolation, which is not terribly convenient, so you may like to try one or more of the many calculators that are aimed to help the designer relate current, conductor width, and conductor thickness:
Doug Brooks: http://www.ultracad.com/calc.htm
George Patrick: http://www.aracnet.com/cgi-usr/gpatrick/trace.pl
Whilst the graphs seem fairly clear in their recommendations, practical experience has indicated that the results they yield are far from accurate. Fortunately conservative, they prove to be based on very old experiments indeed, even before the days of multilayer, so that their extension for use with internal layers was based on pure hypothesis! Nevertheless, they are still the official IPC position. Until, that is, when the new standards, more closely aligned to experimental results, are released. Information on this has been provided for our students by Coretec, and it may be accessed at this link.
But what do you do in a high-power application, when the calculator suggests that, to avoid over-heating and possible damage, a wider track should be used than can be fitted into the space available? The main alternatives are either to select a laminate with a thick copper foil or to increase the track thickness locally by selective plating. In some high-current cases, it may be advantageous to fit bus-bars as components: bear in mind that resistive heating is associated with voltage drop, and this may also impact on circuit performance.
Ground and power connections are usually planes, complete layers of copper connected either to ground or supply voltage. This minimises both the resistance and the inductance of the connection, although the plane is less than perfect because it is pierced with multiple holes. One problem area is getting high currents into a plane: designers must have a sufficient total cross-sectional area of copper in contact with the connector pin(s), and this depends on their number and diameter, as well as the foil parameters.
Contact resistance points on the board, such as an edge connector, may be coated with thicker gold to provide a low resistance connection which can be made and broken during life. This coating is totally different from the electroless nickel with gold flash used for soldering, that we will discuss in later parts of this unit. It typically consists of electrolytically-deposited hard nickel covered with 1-3 µm of hard gold, so that the design has to allow for electrical connection to be made to these areas during the plating operation. The practicalities of this are beyond the scope of this module, and you should consult your board fabricator.
Your circuit contains a power amplifier module which has supply and earth connections which both need a current-carrying capacity of 10 A. The board is a four-layer laminate made with standard FR-4, and the maximum operating temperature of the circuit is 100°C. What options are available to you for supplying this current?
But a PCB is more than just copper! The remainder of this section gives information about the electrical properties of the laminate, in terms of surface resistivity, dielectric strength and breakdown, dielectric constant and dissipation factor. These last two parameters are of particular consequence for high-frequency operation.
The surface resistivity of a board depends on the material, the condition of the surface, and the relative humidity of the surroundings: typical values for glass-epoxy laminates at 60–70% RH are 10–13 to 10–14 W/sq.3
The effect of humidity on surface resistance of glass-epoxy was measured, starting with 97.5% RH at 400C and decreasing the humidity to 64%. The results, shown in Figure 13, indicate that the surface resistivity decreases logarithmically with an increase in humidity at approximately the rate of one decade per 20% humidity change: roughness and contamination also reduce the surface resistivity very markedly.
However, provided that the surfaces are clean, solder mask coatings enhance surface resistivity both in dry and humid conditions, and also reduce any tendency towards electromigration of the conductor material.
These are measures of the ability of an insulation material to resist the passage of a disruptive discharge produced by an electric stress: dielectric strength is measured through the laminate, with values quoted per unit of laminate thickness; dielectric breakdown is measured between two electrodes inserted in the laminate on set centres perpendicular to the laminations. The test results will vary with the material thickness, the form and size of electrodes, the time for which the voltage is applied, the frequency and wave shape of the voltage, the surrounding medium, and the temperature.
The applied voltages are high, and all tests are run under oil to prevent breakdown occurring through the surrounding air. In the short-time tests, the applied voltage is increased at a uniform rate of 0.5 kV/s; in the step-by-step test, the initial voltage is 50% of the short-time breakdown voltage, and the voltage is increased at 1 minute intervals in increments according to a predetermined schedule.
he test values for dielectric strength vary with the thickness of the material, the form and size of electrodes, the time of application of the voltage, the temperature, the frequency and wave shape of the voltage, and the surrounding medium. In Table 3, the test figures are given in the SI unit, which are volts per metre of laminate thickness – in this case MV, because laminates don’t come metres thick! Perhaps a friendlier way of looking at these figures4 is in the original volts per mil (thousandth of an inch), where 19 MV·m–1 = 500 V/mil.
4 In much of the PCB industry, measurements are quoted in Imperial System units, ‘mil’ being the US term for 0.001in, the distance referred to in the UK as ‘thou’.
Dielectric breakdown testing measures the disruptive discharge between two electrodes inserted in the laminate on 25 mm centres. Both short-time and step-by-step tests (Table 4) are performed perpendicular to the laminations, and measured under oil.
‘Electrical clearance’ refers to the minimum distance separating isolated electrical conductors. Recommendations for this are given in Table 5. The distance is a function of the voltage between them, the environment in which the assembly is to be used, and whether the surfaces are conformally coated:
|DC (peak AC)
(sea level to 3 km)
(above 3 km)
|uncoated board||conformal coating|
|50 V||0.63 mm||0.63 mm||0.38 mm|
|500 V||2.54 mm||12.7 mm||1.51 mm|
|>500 V||0.051 mm/V||0.127 mm/V||0.03 mm/V|
The dielectric constant (or ‘permittivity’) is a measure of the ability of an insulating material to store electrostatic energy. It is a dimensionless quantity, whose value is the ratio of the capacitance of a capacitor with a given dielectric to the capacitance of the same capacitor with air as a dielectric. As Table 6 shows, common laminates have dielectric constants in the range 4.0–5.0. These values are calculated from the capacitance as read on a capacitance bridge, the thickness of the specimen, and the area of the electrodes.
|Material||permittivity (at 1MHz)|
Note that these measurements are averages over the electrode area, and there will be local variations because of the presence of the reinforcement, more noticeably with woven reinforcements than with non-woven types. For this reason, specifications for laminates that are designed for high-frequency use will often put limits on the value spread in dielectric constant.
In an insulating material, the dissipation factor is the ratio of the total power loss in the material to the product of the voltage and current in a capacitor in which the material is a dielectric. As with permittivity, dissipation factor varies with changes in temperature and humidity, and also depends on frequency (Table 7). Measurements should therefore always be taken under standard test conditions.
|100 Hz||4.80||0.009||10 MHz||4.55||0.022|
|1 kHz||4.75||0.012||100 MHz||4.50||0.024|
|10 kHz||4.70||0.015||1 GHz||4.45||0.025|
|100 kHz||4.65||0.018||10 GHz||4.40||0.025|
“It’s only a bit of gash insulating material!” Explain to the technician who said this, the ways in which the electrical properties of your FR-4 laminate would be crucial in making a VHF masthead transmitter module operating at high voltage.
[ back to top ]
Power dissipated by a circuit starts as electrical energy, but becomes heat, and this heat needs to be removed in order that the circuit may continue to operate correctly. Where a substantial proportion of the heat is dissipated by just a few parts, or by a single component such as a microprocessor, the designer will often provide a ‘heat sink’, to conduct the heat to a larger surface that may more effectively be kept cool by transferring energy to an external medium. Both natural and ‘forced’ (fan-assisted) convection are common, using the surrounding air to carry heat away; for some high-power systems, the heat sink may be water-cooled.
Unless the sources of heat are many and distributed widely across the board, such strategies are not normally applied to the smaller generators of heat. For these, the thermal energy will be lost by convection to the surrounding air, and by conduction (mostly through the leads) into the board.
Unfortunately, compared with metals, most laminates are very poor conductors of heat: the approximate thermal conductivities of FR-2 and FR-4 are 0.24 W/(m·K) and 0.27 W/(m·K) respectively, against 390 W/(m·K) for copper.
The consequences of this are that:
If you would like to check this, there is an interesting article and calculator in the library at the CoolingZone magazine site.
Note that providing extra copper for heat sinking should not be done at the expense of having the areas of copper evenly distributed, and of balancing the thicknesses of foil used, so that the board is symmetric about its centre plane. If balance is not achieved, this will lead to internal stresses and board warping.
In the last part we considered both operating temperature and the closely-related glass transition temperature: in this section we are looking at what happens when the laminate is taken above its normal operating range.
At high temperatures, thermoset resins will decompose. There is no specific temperature at which this happens, but degradation of characteristics starts near soldering temperatures, with oxidation of the resins leading to darkening. This is the reason why boards used for profiling reflow processes gradually darken in colour, as do boards that have become trapped in a reflow oven.
From the point of view of reliability, however, perhaps the more critical factor is the loss of adhesion which occurs, leading to delamination within the structure. A test procedure that is often quoted is the ‘time to delamination’ test. This measures how long a material will resist blistering or delamination, using a thermomechanical analyser in which the sample is heated to a specified temperature. The most common temperature used has been 260°C (the T260 test), but the higher T288 test is becoming more common with high Tg materials intended for operation with lead-free solders.
Table 8 gives the T260 time for a range of different materials. Broadly speaking the T260 time to blistering/delamination increases with Tg, but there are many differences both between materials and between the same material from different manufacturers.
|material||Tg||T260 (min)||decomposition temperature
(°C for 5% weight loss)
The loss of adhesion associated with delamination and blistering also affects the adhesion between foil and laminate, and at high temperatures the adhesion of foil to the laminate reduces. Such delamination is a frequent occurrence during rework processes, especially when soldering irons are used.
Time to delamination measures the bond between the components of the laminate: in order to assess the actual physical degradation of the resin system that is indicated by colour changes, we use thermogravimetric analysis (TGA). This measures the mass of a sample as it is gradually heated and decomposes. The decomposition temperature reported is the temperature at which 5% of the mass of the sample has been lost through decomposition. Table 8 also gives indicative values for decomposition temperatures: again there is both a relationship between glass transition temperature and decomposition temperature, and some spread of results.
Many of the standards in this area result from concerns from the insurance industry about the consequential damage caused by electrical fires, and the work carried out to qualify materials by Underwriters Laboratories in the USA. The key factors for UL’s sponsors are that a material should self-extinguish (that is, not keep any fire alight), and should not spread the fire by means of flaming droplets.
The UL94 test standard is a generally-used indicator of the acceptability of plastic for general use with regard to its flammability. There are three different tests, and several different levels, as indicated schematically in Figure 14 and Table 9.
The testing for the 94V-0 rating involves two separate applications of a flame for 10s to each sample from two sets of five specimens, and meeting the criteria that:
Most professional laminate materials, such as FR-2 and FR-4, have the highest (94V0) classification for vertical burn tests.
Note that the materials are allowed to burn when the flame is applied, but not to spread a fire (no ‘flaming particles’) or to support continued combustion (must be ‘self-extinguishing’).
There are similar methods in the IPC-TM-650 Test Methods Manual: Methods 2.3.9 for prepreg and thin laminate, 2.3.10 for the flammability of laminate, and 220.127.116.11 for the flammability of solder mask on printed wiring laminate. The tests are comparable, but there are some interesting observations which help shed light on this topic:
Despite the fact that the IPC test is better targeted at the requirements of the PCB industry, UL wins hands down in terms of universal acceptance, because the specification applies to so many of the materials used for purposes such as enclosures.
Provided that sufficient copper had been etched away, it used to be straightforward to see the classification of a laminate from the marking that appeared on one of the inner layers of the board, as a UL logo in combination with the manufacturer’s logo. A red logo indicated that the material was classified UL94 V-0, so the laminate was likely to be FR-2 or FR-4; a blue logo corresponded to a UL94 HB rating, which would indicate either XPC (phenolic) or G10 (glass-epoxy). Unfortunately, this useful feature has more or less disappeared from present-day laminates.
|Surface Burn tests|
Burning must stop within 60 seconds after five applications of five seconds each of a flame to a test bar, and there must be no burn-through hole. The flame is larger than that used in Vertical Burn testing.
This is the highest (most flame retardant) UL94 rating.
|UL94 5VB||As for 94 5VA, but a burn-through hole is allowed.|
|Vertical Burn tests|
|UL94 V-0||Specimens must extinguish within 10s after each flame application and a total combustion of less than 10s after 10 flame applications. No samples are to drip flaming particles or have glowing combustion lasting beyond 30s after the second flame test.|
|UL94 V-1||Specimens must extinguish within 30s after each flame application and a total combustion of less than 250s after 10 flame applications. No samples are to drip flaming particles or have glowing combustion lasting beyond 60s after the second flame test.|
|UL94 V-2||Specimens must extinguish within 30s after each flame application and a total combustion of less than 250s after 10 flame applications. Samples may drip flame particles, burning briefly; and no specimen will have glowing combustion beyond 60s after the second flame test.|
|UL94 VTM||Thin material version of the vertical burning test applied to thin or flexible materials which may distort, shrink or flex during the 94V test. A specimen 200 × 50mm is rolled longitudinally around a 12.7 mm diameter mandrel and taped on one end. When the mandrel is removed the specimen forms a cone shape, which gives it longitudinal rigidity.
Has the same three classifications as 94 V. Differences are that a flame is applied twice for only three seconds, and no specimens may have flaming or glowing combustion beyond a point 125 mm from the bottom of the specimen.
|Horizontal Burn test|
Slow horizontal burning on a 3 mm thick specimen with a burning rate is less than 75 mm/min or stops burning before a mark 125 mm away from the point of flame application. HB rated materials are considered ‘self-extinguishing’.
This is the lowest (least flame retardant) UL94 rating
Your Purchasing Manager has noticed that the paperwork relating to the laminate he/she is now buying refers to compliance with UL94 HB, whereas with most of the material your company uses the reference is to UL94 V-0. Explain the significance of the difference, and the reason for the restricted uses to which the new material can be put.
So far, we have considered the physical, thermal, mechanical and electrical characteristics of a laminate. Before reading further, make as complete a list as possible of any other aspects that you would expect to be covered in a specification for a laminate material.
Review your answer as you read further.
[ back to top ]
As you have discovered so far, laminate materials will be supplied to a specification that includes physical, mechanical, and thermal characteristics, and dielectric parameters such as surface and volume resistivity, dielectric constant and dissipation factor. All these parameters will be fairly generic to the resin/reinforcement selected.
But specifications will also contain much quality and reliability information. In particular, they will give details about the foil that has been used in building the laminate. In the typical commercial foil specification which is given in Table 10, notice that much is made of the way in which the foil is suitable for use both with different materials and different processes. Similar documents from other sources will stress aspects such as bond strength with high Tg resin systems and chemical compatibility.
|minimum adhesion strength of 1 oz. copper at room temperature on conventional epoxy with 130–140°C Tg||9.5 lb/in = 1.66 N/mm|
|high temperature oxidation resistance||180°C for at least 60 min|
|solderability||passes IPC specification|
|UV-curable resist ink adhesion||excellent|
|solder resist ink adhesion||excellent|
|etchability||excellent with cupric chloride, ferric chloride, alkaline, and ammonium persulphate|
|resistance against laminate staining and microvoids||excellent|
|solder float test (288°C, 10 seconds)||<2% adhesion strength loss|
|thermal ageing test (150°C, 10 days)||<15% adhesion strength loss|
Even though the adhesion between foil and laminate also depends on factors other than the foil itself, adhesion strength is a key aspect of the specification. The emphasis reflects both the importance of the parameter and the fact that the surface finish7 and treatment of the foil strongly affect the outcome.
The standard test for adhesion is to measure the resistance exerted on the test equipment when a long strip of foil of defined width is gradually peeled from a laminate. The value quoted will be the minimum load applied divided by the width of the foil strip. The test is defined in IPC-TM-650 Method 2.4.8, and there is also a ‘keyhole method’ variant (Method 18.104.22.168) for thin laminates. Because the measured peel strength is thickness-dependent, tests are normally carried out on 1 oz (35 µm) foil.
Tests are carried out not only on laminate ‘as received’, but also after tests that simulate storage in different conditions, treatment with solvents and simulated electroplating treatment. Usually results8 are little affected by exposure to process chemicals, but deteriorate slightly after thermal stressing (such as floating on molten solder), and reduce by up to 30% at elevated temperatures close to Tg.
8 See Table 8.4 in Section 8.2.6 of Coombs 2001. As always, you can expect to come
across both metric and imperial units. As a guide, 11–12 lb/in = 1.93–2.10 N/mm.
The Z-axis expansion of a laminate increases substantially above Tg, so stresses from changes in temperature can lead to fatigue failure and in consequence the plated layer in the through-holes needs sufficient ductility. However, the foil itself needs reasonable ductility to withstand CTE mismatch, especially where it is constrained from moving, as in the vicinity of a through-hole. Stress on the rest of the inner layer away from the hole can cause a fracture of the copper layer within the board. This is why high-temperature elongation (HTE) foil was developed in the early 1980s. HTE foil has higher elongation at the temperatures typical of soldering processes (Table 11), and its use for inner layers is particularly important in maintaining yields for complex multilayer circuits.
|description||unit||product name||typical values|
|½ oz||1 oz||2 oz|
|tensile strength (RT)||MPa||JTC-HTG||360||325||310|
As you will perhaps have noticed when reviewing the properties of other metals, tensile strength and elongation are inversely correlated, and both parameters depend on the crystal structure, and hence on the method of manufacture and subsequent heat treatment. Table 12 illustrates how ductile a typical electrodeposited (‘Type A’) copper foil can be compared to a rolled (‘Type B’) foil, until that rolled foil has been annealed. The rolled foil will typically only be used for making flexible circuits (a topic we will return to in Technology awareness).
|mass per unit area||minimum tensile strength||minimum elongation percentage|
|Type A (electrodeposited) copper foil|
|152||0.5||10 500||15 000||2||5|
|305||1.0||21 000||30 000||3||10|
|610||2.0||21 000||30 000||3||15|
|Type B (rolled) copper foil, as rolled|
|152||0.5||35 000||50 000||0.5|
|305||1.0||35 000||50 000||0.5|
|610||2.0||35 000||50 000||1|
|Type B (rolled) copper foil, rolled and annealed|
|152||0.5||10 500||15 000||5|
|305||1.0||17 500||25 000||5|
|610||2.0||17 500||25 000||10|
Whilst foil thicknesses are often quoted, foil is usually specified in terms of nominal mass per unit area and percentage deviation from that nominal (typically ±5% or ±10%), with the thickness and thickness deviation given for information only. User preference for the units used in specifications differs between the USA (mostly ounces per square foot) and Europe (mostly grams per square metre), though even in Europe the ½ ounce/1 ounce/2 ounce measure is often to be heard, as it is very convenient.
The inherent conductivity of the foil material may be quoted in absolute SI units (µW m) or relative to the purest possible copper, expressed as the International Annealed Copper Standard (IACS). The BS lower limit for ½ ounce foil is 0.0184 µW or 93.7% IACS, and 0.0179 µW or 96.6% IACS for 1 ounce foil. Such differences in characteristics between thin and thicker foils are to be expected: in the former, a greater percentage of their volume is affected by the material at the surface.
More usually, the specification will quote the maximum resistance of the foil (Table 13). The tests used, whether IEC or IPC, all use a 150 mm section of a longer 25 mm wide foil. The resistance measurements employ four-terminal techniques, because of the low ohmic values, and are corrected to 20°C.
|mass per unit area||maximum resistance|
|152 g/m2 (½ oz/ft2)||7.0 mW|
|305 g/m2 (1 oz/ft2)||3.5 mW|
|610 g/m2 (2 oz/ft2)||1.75 mW|
In order to reach these low resistance values, the copper used must be pure, and some specifications will put limits on the purity of the material. BS EN 60249-3A, for example, demands a minimum purity of 99.8% copper for electrodeposited foils and a minimum of 99.9% for rolled foils. In each case, any silver content is regarded as copper, and the percentage excludes any ‘as shipped’ surface treatment.
Mechanically, typical foil specifications will set standards for the maximum surface roughness, and ask that the foil surfaces should be free of:
The specifications for laminates are similar to those for foil, with limits for inclusions, indentations, bumps, wrinkles and blisters, but with the additional requirement that the surface shall be free of resin, as laminate processing may cause this type of defect. On surface marks, they are arguably less severe: for example, BS EN 60249-2-5 (for FR-4) allows scratches of depth up to 10µm, or 20% of the nominal thickness of the copper foil, whichever is the lower. However, there is an added limitation of 1m per square metre for the total length of scratches deeper than 5µm. Care in handling is needed at every point during laminate manufacture and use!
Under the surface, with the copper etched away, the base material has to be substantially free from pits, holes, scratches, porosity and foreign inclusions (including pre-cured resin particles). Although a small amount of irregular variation of colour is permissible, the base material also has to appear substantially uniform in colour.
The solderability of both copper foil and laminate is also important. Typically the tests carried out evaluate both wetting and de-wetting properties of the foil by simulating wave soldering, rather than using a wetting balance. Samples are first fluxed and then moved at constant speed so that the test face makes contact with the molten solder.
Most of the mechanical aspects have already been considered in the earlier sections . However, laminate specifications will also set limits for dimensional tolerances, including thickness, and for surface waviness (typically ~5µm) as well as warp and twist.
Dimensional stability is important, and this is addressed in BS EN 60249-2-5 by using the test given in BS EN 60249-1 Method 3.11, in which board dimensions are measured before and after exposure to the category temperature for 30 mins, followed by recovery. For FR-4 laminate, the test temperature is 150 ±2°C, and the requirement is that the change in dimension should be less than 0.3mm/m for boards thicker than 0.8 mm (0.5mm/m for thinner boards).
[ back to top ]
The EN specifications we referred to in the previous part are either for foils or for finished laminates, where there is no requirement for specifying the construction of the laminate (which needs to be negotiated with the vendor). IPC-4101A Specification for Base Materials for Rigid and Multilayer Printed Boards covers both finished laminates and prepreg. A large number of ‘slash sheets’ attached to the basic document spell out the specific properties for different classes of materials.
Of course, when we buy prepreg there are subtly different considerations to be borne in mind. Here the prime information required is:
Having a standardised test for the last of these is important, even though the actual flow in any situation will depend on factors such as the bonding between the many different layers. The fabricator must have consistent flow characteristics in order to ensure repeatable manufacturing performance.
The standard four-ply flow test (Test MF) involves cutting pieces of prepreg to a specific size, weighing them and pressing them at a controlled temperature and pressure. The resin flows under the test conditions and afterwards a smaller section is cut from the centre, reweighed and normalized to the original area. The ‘flow' is calculated as the loss in weight between the original material and the final weighed ‘flow biscuit’, expressed as a percentage.
A rose by any other name . . .
So far we have mentioned three standards authorities, so perhaps it is time to look at their role in Electronic Design Realisation.
What are standards?
“Standards are documented agreements containing technical specifications or other precise criteria to be used consistently as rules, guidelines, or definitions of characteristics, to ensure that materials, products, processes and services are fit for their purpose.
For example, the format of the credit cards, phone cards, and ‘smart’ cards that have become commonplace is derived from an ISO International Standard. Adhering to the standard, which defines such features as an optimal thickness (0,76 mm), means that the cards can be used worldwide.
International Standards thus contribute to making life simpler, and to increasing the reliability and effectiveness of the goods and services we use.”
from the ISO web site
So, who is making life simpler? Regrettably too many groups, as Table 14 demonstrates!
ANSI American National Standards Institute
BS EN standards are adopted by ISO, the EN standing for EuroNorm
BS(I) British Standards (Institution)
CECC CENELEC Electronic Components Committee
ENELEC (Comité Européen Normalisation Electrotechnique) European Committee for Electrotechnical Standardisation
IPC Founded as Institute for Printed Circuits, then Institute for Interconnecting and Packaging Electronics Circuits, and now IPC with an accompanying ‘identity statement’ as ‘IPC – Association Connecting Electronics Industries’
ISO International Organization for Standardization9
JIS Japanese Industrial Standard
NEMA National Electrical Manufacturers Association, producers of one of the first classifications for printed circuit boards
9 ISO is a worldwide federation of national standards bodies from more than 140 countries. If you want to know why it isn’t called IOS, visit http://www.iso.ch/iso/en/aboutiso/introduction/whatisISO.html.
When it comes to specifying laminate, you have to be aware that:
For the two materials we have reviewed during this document, Table 15 gives some alternative descriptions. The NEMA and BS 4584 type numbers may technically be obsolete, but you will still find them used, especially the original NEMA names . . .
|NEMA||IPC 4101A||BS 4584||BS EN 60249 Part 2||JIS|
|full form||short form|
[ back to top ]