- Email: [email protected]
Numerical modelling of the heat and mass transport processes in a vacuum vapour phase soldering system
International Journal of Heat and Mass Transfer 114 (2017) 613–620
Contents lists available at ScienceDirect
International Journal of Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ijhmt
Numerical modelling of the heat and mass transport processes in a vacuum vapour phase soldering system Balázs Illés a,c,⇑, Agata Skwarek b, Attila Géczy a,c, Olivér Krammer a,c, David Bušek c a
Department of Electronics Technology, Budapest University of Technology and Economics, Budapest, Hungary Department of Microelectronics, Institute of Electron Technology, Krakow, Poland c Department of Electrotechnology, Czech Technical University in Prague, Czech Republic b
a r t i c l e
i n f o
Article history: Received 7 May 2017 Received in revised form 10 June 2017 Accepted 21 June 2017
Keywords: Vacuum vapour phase soldering Standard k-e turbulence method RANS Vapour concentration Gas void
a b s t r a c t The heat and mass transport processes were investigated with numerical simulations in a vacuum vapour phase soldering system during the vapour suctioning process. Low vapour pressure/concentration is applied during vacuum soldering to decrease the number of gas voids in the solder joints. Threedimensional numerical flow model was developed which based on the Reynolds averaged NavierStokes equations with the standard k-e turbulence method. The decrease of the vapour concentration and its effects on the solder joints were studied in the case of different oven settings. It was found that vapour suctioning has considerable effects on the heat transfer processes in the soldering chamber which might lead to early solidification of the solder joints and reduces the efficiency of the void removal. Different oven settings were simulated in order to decrease the heat loss of the soldering chamber during the vapour suctioning. It was shown that with appropriate setting of the vacuum vapour phase soldering technology, the efficiency of the void removal can be increased. Ó 2017 Elsevier Ltd. All rights reserved.
1. Introduction The physical phenomenon of condensation is widely used for heating and cooling technologies due to its high efficiency, such as facility heating with heat pumps [1], cooling refrigerators with hydro-chlorofluorocarbons (HFC) refrigerants [2], or microelectronics with heat pipes [3]. The Vapour Phase Soldering (VPS) or ‘‘condensation soldering” is a reflow soldering method. It is considered as a promising alternative of convection and infrared reflow methods [4,5]. Before the reflow soldering process, the solder paste is deposited onto the solder pads of a Printed Circuit Board (PCB) (Fig. 1a) with stencil printing (Fig. 1b). Then, discrete components of the circuits are placed onto solder deposits (Fig. 1c). Finally, the assembly is heated up over the melting point of the applied solder alloy which forms mechanical and electrical joint between the terminals of the components and pads of the PCB. The principle of vapour phase soldering is using the heat transfer effect of condensation. During the process a special heat transfer fluid is heated at the bottom of a tank. When the fluid is heated up to its boiling point, a vapour space begins developing which fill up a closed tank (Fig. 1d). The excessive vapour is condensed on ⇑ Corresponding author at: Department of Electronics Technology, Budapest University of Technology and Economics, Budapest, Hungary. E-mail address: [email protected] (B. Illés). http://dx.doi.org/10.1016/j.ijheatmasstransfer.2017.06.091 0017-9310/Ó 2017 Elsevier Ltd. All rights reserved.
the top of the tank, due to a cooling pipe setup. When the vapour space is ready for soldering, the assembled PCB is immersed into the vapour (Fig. 1d), and a condensate layer forms on the colder surface of the PCB (Fig. 1e). This layer transfers the latent heat of condensing mass and the conducted heat from surrounding vapour to the assembly, which is heated up to the boiling point of the heat transfer fluid. After the melting and wetting of the solder alloy, the PCB is lifted out of the process zone in order to cool down, and to solidify the solder joints (Fig. 1f). Nowadays the most widely applied heat transfer fluid is Galden, which contains ether chains closed with carbon-fluorine bonds (Perfluoropolyether, PFPE). The boiling point of the Galden liquid can be set with the length of the ether chain between 150 and 260 °C [7]. The main advantages of condensation heating for soldering are lack of overheating [8] and lack of shadowing effect, which occurs due to larger components [4]. However, intensive heat transfer during VPS and the hermetically closed process zone can also cause soldering failures like voiding, paste sputtering, tombstone failures [9], since the heat transfer coefficient of Galden vapour can be 2–3 times higher than the heat transfer coefficient of gas streams in a convection oven [10]. Up to now, only limited researchers have dealt with thermal aspects of the process. Leicht et al. showed that the heat transfer coefficient of vapour can be decreased with the application of non-saturated vapour
614
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
SMT PROCESS
a.)
STANDARD VPS PROCESS
Immersion
d.)
Assembled SMT PCB Vapour Galden
Printed Circuit Board (PCB) prepared for assembly
b.)
Stencil foil
Heaters
e.)
Paste deposit
Solder paste printing with stencil technology
Lift up
c.) Placing machine Auto component placement (assembled SMT PCB)
f.)
Dwelling in saturated vapour (heating)
Removal PCB, cooling
Fig. 1. The SMT and the basic VPS process [6].
[11]. Dumitru et al. investigated the effect of heating of VPS process on the mechanical characteristics of PCBs [12]. Cucu et al. studied the mechanical stability of the solder joint prepared by VPS process from low melting point alloys [13]. Krammer showed that a proper VPS thermal profile can increase the reliability of solder joints [14] since both the thickness and the shape of the intermetallic layer will differ from the conventional soldering technology. Synkiewicz et al. presented similar results about the general quality of the solder joints for thermo-generators [15]. Géczy et al. optimized temperature profiling methods for VPS process [16]; they also proved with pressure measurements that temperature saturation of the process zone occurs before vapour concentration saturation in time [17]. It was shown that condensate thickness changes considerably on the surface of the PCB which results in spatial heat transfer differences. In addition it was proven that the dense vapour space takes place considerably (20%) in the heating of the assembly [18,19]. The gas voids in the joints causing serious reliability problems since it decrease the joints life time and electrical conductivity. Krammer et al. proved that one of the key points to prevent of void formation is the appropriate amount of solder paste [20]. However, voids can even form during the stencil printing, which highly depends on the viscosity of the solder paste [21]. The issue of voiding was thoroughly investigated in the case of VPS and the problem was found even more serious than in the case of conventional soldering technologies [9]. Therefore, studies have started about the possible application of the low pressure for VPS system, where the voids are removed from the molten solder [22]. The suction process takes place while the alloy is above its melting point [22]. The pressure in the ‘‘vacuum chamber” is usually approximately 40–50 mbar during the suction part of the process [23]. However it was never studied that how does the decrease in vapour concentration affect the heat transport mechanism of the soldering process, and with this aspect, how does it affect the efficiency of the void removal and the forming of the solder joints microstructure. The question is relevant from the aspect of mass
manufacturing, while vacuum type ovens recently started to spread out in electronics assembly plants. 2. Numerical modelling of vacuum VPS process Investigating the VPS process by numerical simulations is necessary because of the closed environment (high temperature, dense vapour, etc.) in the soldering chamber. There were experimental studies about application of the temperature, pressure or optical probes in VPS ovens [16,24,25] but it was found that the introduction of the probes is complicated even in the case of simple VPS ovens. In the case of vacuum VPS process when the soldering chamber is hermetically closed, the measurement inside the soldering chamber is not possible with such probes which need to be connected to an outer device (like thermocouples, pressure sensors, borescopes, etc.). 2.1. Physical description of the model According to the preliminary calculations, the vapour concentration decreasing in the vacuum chamber causes a turbulent gas flowing [26], so the numerical calculations are based on the 3 dimensional Reynolds Averaged Navier-Stokes (RANS) equations. The RANS equations compute the average movement in a turbulent flow, while the effect of fluctuation is modelled by Reynold’s stress tensor which was related to the mean flow with a standard k-e turbulence model [27]. The decrease of the vapour concentration is initialized by a pressure drop generated by a vacuum pump. The governing equations of the vapour space are the following: the transient continuity equation for compressible Newtonian fluids is used since the local pressure change has effect on the density of the vapour:
@q @ui ¼0 þq @t @xi
ð1Þ
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
The RANS equation for compressible Newtonian fluids is used to close governing equations. Here it is formulated by the Einstein notation:
@ui @ui 1 @p @ 2 @uj ¼ gi þ ðt þ tT Þ 2Sij dij þ uj 3 @xj @t @xj q @xi @xj
ð2Þ
615
ing mass is determined by the heat flux which the condenser surface (in the case the suction pipe wall) can conduct away [29]:
@mc kðpÞ AðpÞ @T ðpÞ ¼ @t h @xj
ð10Þ
where A(p) is the area of the pipe wall [m2].
where the mean rate of strain tensor is:
Sij ¼
1 @ui @uj þ 2 @xj @xi
2.2. Numerical solution and parameters of the model
ð3Þ
where k is the turbulent kinetic energy:
The numerical conversion of the previously defined partial differential equations was done by Finite Difference Method (FDM) and was solved by explicit Forward Time Central Space (FTCS) algorithm in order to achieve high solving calculation speed and flexible implementation. The general numerical form of FTCS is the following:
@k @k ¼ þ uj @t @xj
@v @v @v 2 ¼0! þ þ FTCS @t @x @ 2 x
where the turbulent eddy viscosity is:
tT ¼ C l
k
2
ð4Þ
e 2 @uj 2 @ui 1 @ q @p tT 2Sij dij kdij tT 3 3 @xj 3 @xj q @xi @xi @ t @k eþ tþ T @xj rk @xj
ð5Þ
@e @e ¼ C e1 þ uj @t @xj
2 @uj 2 @ui 1 @ q @p tT 2Sij dij kdij tT 3 3 @xj 3 k @xj q @xi @xi e2 @ tT @ e tþ C e2 þ k @xj re @xj
e
ð6Þ The applied constants were the following (these are the generally used constants of the standard k-e model [28]):
C l ¼ 0:09;
rk ¼ 1; re ¼ 1:3; C e1 ¼ 1:44; C e2 ¼ 1:92;
The applied markings are the following: u is the velocity [m/s], m is the kinematic viscosity [m2/s] of the Galden vapour, p is the pressure [Pa], q is the concentration of the Galden vapour [kg/ m3], g is the gravity acceleration [m/s2] and dij is the Kronecker delta. The vapour pressure is calculated according to the general gas law:
p¼
amuH qRT amuGalden
ð7Þ
where amuH is the average molecular weight of hydrogen, amuGalden is the average molecular weight of Galden, R is the gas constant (8.314 J K1 mol1) and T is the temperature [K]. Turbulent transport of energy is also described by the averaging method:
@T @T k @ @T þ uj ¼ @t @xj qC s @xj @xj
ð8Þ
where k is the specific thermal conductivity [W/m K] and Cs is the specific heat capacity [J/kg K]. During the suctioning process, the vapour will condensate on the colder wall of the suction pipe. In this application, the formation of the condensate layer on the suction pipe wall and the presumed forming of dew point [19] at the end of the suction process are neglected. Therefore, the heat transfer by the condensation can be calculated only with the latent heat of the condensing mass [29]:
@T ðpÞ h @mc ¼ C ðpÞ @t @t
v
k iþ1
¼0
and e is the turbulent dissipation rate:
þ
v kþ1 v ki v kiþ1 uki1 i
ð9Þ
where T(p) is the pipe temperature [K], C(p) is the heat capacity of the pipe [J/K], h is the latent heat of the Galden [J/kg] and mc is the amount of condensing of Galden [kg]. The amount of the condens-
þ Dt 2 Dx k 2v i þ v ki1 Dx2
ð11Þ
where v is a general variable, t is the time [s], x is the space [m] and i and k are the indexing of the numerical grid and time steps respectively. The numerical model was built and solved in MATLAB software. A commercial vacuum VPS system was studied. The working principle of the oven is the following: until the melting of solder paste, it works like a simple VPS oven with a non-hermetic soldering chamber. After the melting of the solder alloy a cap closes the soldering chamber hermetically and forming a vacuum chamber with 325 325 85 mm dimensions. A suction pipe is located at the middle top of the closing cap with 13 mm inner and 24 mm outer diameter made from silicone rubber. The vacuum pump of the oven works with ‘‘volume separation” method which means that the pump separates the same volume of vapour out from the workspace in a given time. This implies constant decrease in the relative pressure between the suction pipe and the vacuum chamber. The vacuum pump decreases the pressure to 5 kPa in 5 s and the assembly spends further 15 s at this state to give enough time to the voids to leave the solder joints. (No more details or data about the working principle of the vacuum pump is known.). The possible effects of the heat transport changes during the vapour suctioning was studied on a FR4 Printed Circuit Board (PCB) with soldered resistors (with Al2O3 ceramic body). SAC305 type solder alloy was applied in 125 mm thickness (standard stencil thickness for solder paste printing) on the 17 mm thick (standard solder pad thickness) Cu solder pad. The surface of the PCB was 160 160 mm2 and 3 different thicknesses (0.5, 1 and 2 mm) were examined. The structure of the vacuum chamber with the assembly and the detailed structure of the assembly can be seen in Fig. 2. LS230 type Galden was used for the investigations. The physical parameters of the different materials can be seen in Table 1. The cell geometries and the time step was set by the Courant– Friedrichs–Lewy (CFL) condition, which has to be applied for all of the cells in the model:
Dt
ux uy uz 61 þ þ Dx Dy Dz
ð12Þ
According to the CFL condition, the soldering chamber was divided to 48,000 cells and the time step was set to 105 s. The assembly was divided to 2000 cells. The assembly model was embedded into the vacuum chamber model and it runs as a cosimulation (details about this modelling technique can be read in
616
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
Suction pipe Vacuum chamber Assembly
SAC305 solder
0.5mm Al2O3 resistor body
17µm thick Cu pad
0.5; 1; 2mm Fr4 board
Fig. 2. The structure of the soldering chamber with the assembly and the detailed structure of the assembly.
Table 1 Physical properties of the applied materials.
a b
Property/material
LS230
FR4
Cu
SAC305
Al2O3
Silicone rubber
Boiling point [°C] Density [kg/m3] Dynamic viscosityb [Ps s] Average molecular weight Specific thermal cond. [W/m k] Specific heat cap. [J/kg K] Latent heat [J/kg]
230 19.96a 2.13 105 1010 0.065 973 6.3 104
– 2100 – – 0.23 570 –
– 8960 – – 385 3850 –
– 7370 – – 58 2320 –
– 3690 – – 18 880 –
– 1420 – 0.35 1200 –
The saturated vapour concentration of the LS230 Galden. The dynamic viscosity of the LS230 Galden vapour was calculated by the method of Lee et al. in [30].
[19]). Due to the symmetry of the oven only a quarter of the whole geometry was calculated. At the start of the vapour suctioning, saturated vapour conditions were applied as an initial conditions. Saturated vapour condition was applied for the model as initial condition which means: q(0) = 19.96 kg/m3, T(0) = 230 °C, and the vapour is not moving before the suctioning, v(0)=0 m/s. Since the soldering chamber is not hermetically closed before the suctioning process the initial condition of the Galden vapour pressure is p(0) = 101 kPa. The initial temperature of the assembly is also 230 °C. Initial condition of the turbulent kinetic energy:
3 k ¼ ðU IÞ2 2
ð13Þ
where U is the mean flow velocity and I is the turbulence intensity which was chosen to 0.1 (the flow is supposed to be highly turbulent). Initial condition of the turbulent dissipation rate: 3
e ¼ Cl
k2 l
ð14Þ
where l is the turbulent length scale which is 0.1 m in this case. During the preliminary investigations it was found that during the suctioning process the soldering chambers suffers a heat loss on the suction pipe. The initial temperature of the suction pipe
and its change during the suctioning was measured with K-type thermocouples. It was found that initial temperature of the suction pipe is Tp(0) = 60 °C which increases 1.6 °C during the 5 s of suctioning. Fig. 3 shows a proper fitting between the calculated and measured temperatures of the suction pipe during the suctioning process. Boundary conditions of the model are the followings: the walls of the soldering chamber are adiabatic and the flow velocity tends to zero at the walls:
@T ¼ 0 @r r!wall
and
lim
v ¼0
r!wall
ð15Þ
3. Results and discussion The first step of the investigations was to determine the relative pressure decrease generated by the vacuum pump which is necessary to reach the 5 kPa pressure of the soldering chamber in 5 s. During the preliminary calculations with a laminar flow model it was found that this relative pressure decrease is 1.22%p, however the preliminary calculations also showed that according to the Reynolds numbers the flow has to be highly turbulent [26]. The
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
Fig. 3. Calculated and measured temperature of the suction pipe during the suctioning.
turbulence decreases the average flow velocity, therefore it was supposed that in reality the relative pressure decrease of the vac-
617
uum pump has to be higher than 1.22%p. The approximation started from 1.22%p in 0.01 steps. Fig. 4a) shows the pressure and the concentration decrease in the soldering chamber in the case of different vacuum pump settings. The results shows that the -3.4%p vacuum pump setting approximates the real operation, in this case the vapour concentration decreases to 1.98 kg/m3. The ratios of the pressure and concentration decreases are not the same since the soldering chamber suffers a heat loss during the suctioning process. This results in that the pressure decreases to 5% while the vapour concentration only to 10%. The flow velocity analyses showed that at the beginning of the suctioning process (between 0 and 0.075 s) the movement of the vapour is highly fluctuating in the soldering chamber (Fig. 4b and c). This is caused by the fluctuation of the pressure. After 0.075 s till the end of the suctioning process the flow velocities remains constant. At the beginning of the suctioning (0.01 s) the flow velocity reaches 28 m/s at the entrance of the suction pipe, and after a high drop down, it is stabilizing around 14 m/s (Fig. 4b). There are high flow velocity differences diagonally in the soldering chamber, the vapour almost stands at the corners and the flow velocity is still only 2 m/s at the half of the diagonal space (Fig. 4b). Towards the vertical direction (under the suction pipe) the velocity differences are smaller, the flow velocity is 9 m/s at the bottom of the soldering chamber (Fig. 4c). It can
Fig. 4. Vapour space parameters during the suctioning process: (a) decrease of the pressure and vapour concentration in the case of different vacuum pump settings; (b) flow velocities diagonally in the soldering chamber (3.4%p) between 0–0.2 s; (c) flow velocities vertically in the soldering chamber (3.4%p) between 0–0.2 s.
618
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
the top and the bottom level of the soldering chamber the vapour concentration difference is: 1.5 kg/m3 (8.3%) at 0.1 s and 0.1 kg/ m3 (7.1%) at 5 s. According to the energy transport calculations, the heat loss on the suction pipe causes considerable heat loss in the soldering chamber during the suctioning. Fig. 6 shows the temperature changes during the whole void removal process (20 s) in the case of different settings. During the suctioning (0–5 s) the vapour temperature decreases to 80 °C (Fig. 6b). The vapour temperature differences are low during the suctioning (between 2–4 °C). However during the rest of the void removal process (between 5–20 s), the vapour nearby the assembly is heated back (to 130– 145 °C) by the hotter assembly, but this heating effect on the whole chamber is moderate. The observed heat loss and the vapour concentration differences effect on the heat transport processes of the soldering. It is possible that the solder joints can be solidified much earlier than the end of the void removal and the voids cannot leave the solder joints. This effect was studied on solder joints (Fig 2) placed at different positions on different thick PCBs (Fig. 6a). Fig. 6c) shows the minimum and maximum solder joint temperatures on PCB with different thickness during the void removal. During the suctioning the temperature of the solder joints decreases considerably with 8–14 °C. In the cooling of the solder joints the thickness of the PCB is important since this determines the total heat of the assembly. The temperature decrease is the most serious in the middle of the PCB (point 9), the heat loss is less towards the corners of the
be concluded that the suctioning moves the vapour the most in the middle space of the soldering chamber. In the next step the changes of the vapour concentration was studied during the suctioning process at different levels of the soldering chamber. Fig. 5 shows the vapour concentration at the bottom, at the middle and at the top of the soldering chamber. At the beginning of the suctioning process (0–0.03 s), the fluctuation of the flow velocity changes the distribution of the vapour concentration (Fig. 5a and b). Up to 0.1 s the distribution of the vapour concentration is stabilizing and remaining similar till the end of the suctioning (Fig. 5c and d). It was found that after the distribution of the vapour concentration stabilized, considerable concentration gradient can form, mainly horizontally, between the suction pipe and the corners of the vacuum chamber (Fig. 5c). The maximum difference is 2.25 kg/m3 (12%) at the top level of the soldering chamber at 0.1 s. During the suctioning this differences is decreasing slightly, at the end of the process (Fig. 5d), it is 0.15 kg/m3 (10%). Vertically, along the height of the soldering chamber, smaller vapour concentration differences were found than horizontally. Under the suction pipe between the top and the bottom level of the soldering chamber the vapour concentration difference is: 1.5 kg/m3 (8.3%) at 0.1 s and 0.1 kg/m3 (7.1%) at 5 s. During the suctioning this differences is decreasing slightly, at the end of the process (Fig. 5d), it is 0.15 kg/m3 (10%). Vertically, along the height of the soldering chamber, smaller vapour concentration differences were found than horizontally. Under the suction pipe between
b) Vapour concentration [kg/m3]
a) Vapour concentration [kg/m3]
20
Bottom
20.5
Bottom
19.5
20
19
19.5
Middle
19
Middle
18.5 18
18.5
Top Top
18 18
18 12
12
Cham
6
6
ber d
im. [c
m]
0 0
im.
ber d
Cham
17.5 18
[cm]
18 12
Cham12 ber d im
6
. [cm]
6 0 0
. [cm]
im
ber d
Cham
d) Vapour concentration [kg/m3]
c) Vapour concentration [kg/m3]
1.5
19
Bottom
Bottom
18.5
1.45
18
1.4
17.5
1.35
Middle
17
Middle
1.3
Top
Top
16.5 18
18 12
12
Cham
ber d
6
6
im. [c
m]
0 0
Cham
ber d
m] im. [c
1.25 18
18 12
12
Cham
ber dim
6
. [cm]
6
0 0
. [cm]
er dim
b Cham
Fig. 5. Vapour concentration at different levels of the soldering chamber (quarter sections, 3.4%p): (a) at 0.015 s; (b) at 0.03 s; (c) at 0.1 s and (d) at 5 s.
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
619
Fig. 6. Temperature changes in the soldering chamber during the void removal: (a) component positions on the PCB; (b) temperature changes of the vapour; (c) minimum and maximum solder joint temperatures in the case of different PCB thicknesses; (d) minimum solder joint temperatures in the case of different oven settings.
PCB (point 1). This effect is caused by the characteristic of the suctioning, namely that the vapour concentration (Fig. 5c and d) as well as the vapour temperature (Fig. 6a) is the lowest at the middle of the soldering chamber where the assembly is usually positioned. The solidus temperature of the lead free solder alloy is around 219 °C, according to the calculations, if the PCB is thinner than 2 mm, the solder joints can be solidified (or get very close to the solidification) to the end of the suctioning. Although after the suctioning the temperature of the solder joints increases a bit with 2 °C, it is caused by the heating effect of the assembly (the PCB and the ceramic body of the components remain at higher temperature than the solder joints) and the stop of the convection. This can cause that the solidified joints melt again, but the time of the void removal will be still shorter than it has to be. Different oven settings were studied in order to decrease the heat loss during the suctioning process. Fig. 6d) shows the lowest solder joint temperatures on a 1.5 mm thick PCB in the case of different oven settings. The intensity of the vapour concentration decrease has significant effect on the heat loss since it determines the flow velocity and via this the convection. If the relative pressure decrease of the vacuum pump is only 2.5%p (instead of 3.4%p) the positive effect is already visible (Fig. 6d), the temperature of the solder joints remains over 221.5 °C. However the 221.5 °C is still very close to the liquids point and with this vacuum
pump setting the pressure remains 20 kPa in the soldering chamber after 5 s suctioning (Fig. 4a) which ensures less effective void removal than the 5 kPa in the case of 3.4 %p. Since the heat loss is happening on the suctions pipe, the heating of the suction pipe would be an evident solution. If the temperature of the suction pipe would be elevated from 60 to 120 °C, considerable improvement could be reached. The solder joints temperature remains over 223 °C (Fig. 6d) without losing the suction power of the low pressure in the soldering chamber. The thermal diffusivity of the vacuum pipe was also examined, but the thickness increase of the vacuum pipe wall has got only a minor effect on the heat loss. In the last step the calculations were done with a heat transfer fluid with higher boiling point (Galden HS240, boiling point 240 °C). This change showed the best solution, the temperature of the solder joints remains over 229 °C during the whole void removal process (Fig. 6d). 4. Conclusions The heat and mass transport processes was investigated by numerical simulations in a vacuum type VPS oven during the vapour suctioning process. For this purpose a three-dimensional numerical gas flow model was developed which based on the RANS equations with the standard k-e turbulence method. It was found
620
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
that during the vapour suctioning, high flow velocity differences can form in the soldering chamber which causes considerable vapour concentration gradient, mainly horizontally, between the suction pipe and the corners of the vacuum chamber. The oven suffers a heat loss on the suction pipe which causes the drop of the solder joints temperature with 8–14 °C (determined by the position of the given joint on the PCB and the thickness of the PCB). The combined effect of the vapour concentration gradient and the temperature drop can cause the following phenomenon: the solder joints will solidify during the vapour suctioning process before the voids could be removed from the solder joints. This side effect inhibits the further void removal from the solder joints and this has negative effect on the formation, and the overall quality and reliability aspects of the solder joints microstructure. In order to ensure effective void removal, the vacuum VPS technology has to be set/modify carefully (e.g. heating of the suction pipe) mainly in the function of the heat capacity of the soldered assembly. Nevertheless, in the case of common lead-free alloys, the application of heat transfer fluid with the highest suggested boiling point (240 °C) and PCBs having thicknesses 1.5 mm or thicker are suggested for vacuum VPS technology. Conflict of interest The authors declared that there is no conflict of interest. Acknowledgement This research has been partially supported by the ‘‘ÚNKP” program of the Hungarian Government. References [1] C. Naldi, E. Zanchini, Dynamic simulation during summer of a reversible multifunction heat pump with condensation-heat recovery, App. Therm. Eng. 116 (2017) 126–133. [2] J. Kaew-On, N. Naphattharanun, R. Binmud, S. Wongwises, Condensation heat transfer characteristics of R134a flowing inside mini circular and flattened tubes, Int. J. Heat Mass Trans. 102 (2016) 86–97. [3] F.L. Chang, Y.M. Hung, Dielectric liquid pumping flow in optimally operated micro heat pipes, Int. J. Heat Mass Trans. 108/A (2017) 257–270. [4] B. Illés, I. Bakó, Numerical study of the gas flow velocity space in convection reflow oven, Int. J. Heat Mass Transf. 70 (2014), 195-191. [5] B. Kovács, A. Géczy, G. Horváth, I. Hajdu, L. Gál, Advances in Producing Functional Circuits on Biodegradable PCBs, Period. Polytech. Elec. Comp. Sci. 60 (4) (2016) 223–231. [6] B. Illés, A. Géczy, Investigating the heat transfer on the top side of inclined printed circuit boards during vapour phase soldering, App. Therm. Eng. 103 (2016) 1398–1407. [7] R.P. Nielsen, R. Valsecchi, M. Strandgaard, M. Maschietti, Experimental study on fluid phase equilibria of hydroxyl-terminated perfluoropolyether oligomers and supercritical carbon dioxide, J. Supercrit. Fluids 101 (2015) 124–130. [8] J. Huang, J. Zhang, L. Wang, Review of vapor condensation heat and mass transfer in the presence of non-condensable gas, App. Therm. Eng. 89 (2015) 469–484.
[9] B.K. Synkiewicz, A. Skwarek, K. Witek, Voids investigation in solder joints performed with vapour phase soldering (VPS), Solder. Surf. Mount. Tech. 26 (1) (2014) 8–11. [10] H.K. Dawood, H.A. Mohammed, N.A.C. Sidik, K.M. Munisamy, M.A. Wahid, Forced, natural and mixed-convection heat transfer and fluid flow in annulus: a review, Inter. Comm. Heat Mass Transf. 62 (2015) 45–57. [11] H. Leicht, A. Thumm, Today’s Vapor Phase Soldering - An Optimized Reflow Technology for Lead Free Soldering, in: Proceed. of Surf. Mount Tech. Assoc. Intern. Conf., Orlando, USA, paper No. 45, 2008. [12] G.I. Dumitru, A. Tudor, G. Chisiu, I. Plotog, Mechanical characteristics of electronic printed circuit obtained by the vapour phase soldering process, App. Mech. Mater. 658 (2014) 71–76. [13] T.C. Cucu, I. Plotog, M. Branzei, Mechanical tests regarding low-temperature lead-free solder pastes application in automotive electronics, in: Proc. of 20th International Symposium for Design and Technology in Electronic Packaging (SIITME), Bucharest, Romania, 2014, pp. 63–68. [14] O. Krammer, Comparing the reliability and intermetallic layer of solder joints prepared with infrared and vapour phase soldering, Solder. Surf. Mount Technol. 26 (2014) 214–222. [15] B. Synkiewicz, A. Skwarek, K. Witek, Vapour phase soldering used for quality improvement of semiconductor thermogenerators (TEGs) assembly, Mater. Sci. Semicond. Process. 38 (2015) 346–351. [16] A. Géczy, B. Kvanduk, B. Illés, G. Harsányi, Comparative study on proper thermocouple attachment for vapour phase soldering profiling, Solder. Surf. Mount Technol. 28 (2016) 7–12. [17] A. Géczy, B. Illés, Z. Péter, Z. Illyefalvi-Vitéz, Characterization of vapour phase soldering process zone with pressure measurements, Solder. Surf. Mount Technol. 25 (2) (2013) 99–106. [18] B. Illés, A. Géczy, Investigating the dynamic changes of the vapour concentration in a Vapour Phase Soldering oven by simplified condensation modelling, App. Therm. Eng. 59 (2013) 94–100. [19] B. Illés, A. Géczy, Numerical simulation of condensate layer formation during vapour phase soldering, App. Therm. Eng. 70 (2014) 421–429. [20] O. Krammer, B. Varga, K. Dušek, New method for determining correction factors for pin-in-paste solder volumes, Solder. Surf. Mount Technol. 29 (1) (2017) 2–9. [21] O. Krammer, B. Varga, D. Busek, Investigating the effect of solder paste viscosity change on the pressure during stencil printing, in: Proc. of 22nd International Symposium for Design and Technology in Electronic Packaging (SIITME), Oradea, Romania, 2016, pp. 36–39. [22] C. Zabel, A. Duck, Vapour phase vacuum-soldering. A new process technology opens tremendous production capabilities when reflow-soldering, SMTA International Conference, San Diego, California, October 4–8, 2009. [23] C. Zabel, A. Duck, Vapour phase vacuum-soldering. A new process technology opens tremendous production capabilities when reflow-soldering, SMTA International Conference, San Diego, California, October 4–8, 2009. [24] A. Géczy, D. Nagy, B. Illés, L. Fazekas, O. Krammer, D. Busek, Studying heat transfer on inclined printed circuit boards during vapour phase soldering, Solder. Surf. Mount. Tech. 29 (1) (2017) 34–41. [25] L. Livovsky, A. Pietrikova, Real-time profiling of reflow process in VPS chamber, Solder. Surf. Mount. Tech. 29 (1) (2017) 42–48. [26] B. Illés, A. Skwarek, A. Géczy, K. Witek, Numerical Study of Vapour Concentration Reduction in a Vacuum Vapour Phase Soldering Oven, in: Proc. of 40th International Spring Seminar on Electronics Technology, Bulgaria, Sofia, 2017, in press. [27] Y.-Y. Bae, E.-S. Kim, M. Kim, Assessment of low-Reynolds number k-e turbulence models against highly buoyant flows, Int. J. Heat Mass Trans. 108/A (2017) 529–536. [28] H.K. Versteeg, W. Malalasekera, in: An Introduction to Computational Fluid Dynamics: The Finite Volume Method, Chapter 3: Turbulence and its modelling, second ed., Harlow, Pearson Education, 2007, pp. 42–65. [29] B. Illés, A. Géczy, A. Skwarek, D. Busek, Effects of substrate thermal properties on the heat transfer coefficient of vapour phase soldering, Int. J. Heat Mass Trans. 101 (2016) 69–75. [30] A.L. Lee, M.H. Gonzalez, B.E. Eakin, The viscosity of natural gases, J. Pet Technol. 18 (8) (1966) 997–1000.
Contents lists available at ScienceDirect
International Journal of Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ijhmt
Numerical modelling of the heat and mass transport processes in a vacuum vapour phase soldering system Balázs Illés a,c,⇑, Agata Skwarek b, Attila Géczy a,c, Olivér Krammer a,c, David Bušek c a
Department of Electronics Technology, Budapest University of Technology and Economics, Budapest, Hungary Department of Microelectronics, Institute of Electron Technology, Krakow, Poland c Department of Electrotechnology, Czech Technical University in Prague, Czech Republic b
a r t i c l e
i n f o
Article history: Received 7 May 2017 Received in revised form 10 June 2017 Accepted 21 June 2017
Keywords: Vacuum vapour phase soldering Standard k-e turbulence method RANS Vapour concentration Gas void
a b s t r a c t The heat and mass transport processes were investigated with numerical simulations in a vacuum vapour phase soldering system during the vapour suctioning process. Low vapour pressure/concentration is applied during vacuum soldering to decrease the number of gas voids in the solder joints. Threedimensional numerical flow model was developed which based on the Reynolds averaged NavierStokes equations with the standard k-e turbulence method. The decrease of the vapour concentration and its effects on the solder joints were studied in the case of different oven settings. It was found that vapour suctioning has considerable effects on the heat transfer processes in the soldering chamber which might lead to early solidification of the solder joints and reduces the efficiency of the void removal. Different oven settings were simulated in order to decrease the heat loss of the soldering chamber during the vapour suctioning. It was shown that with appropriate setting of the vacuum vapour phase soldering technology, the efficiency of the void removal can be increased. Ó 2017 Elsevier Ltd. All rights reserved.
1. Introduction The physical phenomenon of condensation is widely used for heating and cooling technologies due to its high efficiency, such as facility heating with heat pumps [1], cooling refrigerators with hydro-chlorofluorocarbons (HFC) refrigerants [2], or microelectronics with heat pipes [3]. The Vapour Phase Soldering (VPS) or ‘‘condensation soldering” is a reflow soldering method. It is considered as a promising alternative of convection and infrared reflow methods [4,5]. Before the reflow soldering process, the solder paste is deposited onto the solder pads of a Printed Circuit Board (PCB) (Fig. 1a) with stencil printing (Fig. 1b). Then, discrete components of the circuits are placed onto solder deposits (Fig. 1c). Finally, the assembly is heated up over the melting point of the applied solder alloy which forms mechanical and electrical joint between the terminals of the components and pads of the PCB. The principle of vapour phase soldering is using the heat transfer effect of condensation. During the process a special heat transfer fluid is heated at the bottom of a tank. When the fluid is heated up to its boiling point, a vapour space begins developing which fill up a closed tank (Fig. 1d). The excessive vapour is condensed on ⇑ Corresponding author at: Department of Electronics Technology, Budapest University of Technology and Economics, Budapest, Hungary. E-mail address: [email protected] (B. Illés). http://dx.doi.org/10.1016/j.ijheatmasstransfer.2017.06.091 0017-9310/Ó 2017 Elsevier Ltd. All rights reserved.
the top of the tank, due to a cooling pipe setup. When the vapour space is ready for soldering, the assembled PCB is immersed into the vapour (Fig. 1d), and a condensate layer forms on the colder surface of the PCB (Fig. 1e). This layer transfers the latent heat of condensing mass and the conducted heat from surrounding vapour to the assembly, which is heated up to the boiling point of the heat transfer fluid. After the melting and wetting of the solder alloy, the PCB is lifted out of the process zone in order to cool down, and to solidify the solder joints (Fig. 1f). Nowadays the most widely applied heat transfer fluid is Galden, which contains ether chains closed with carbon-fluorine bonds (Perfluoropolyether, PFPE). The boiling point of the Galden liquid can be set with the length of the ether chain between 150 and 260 °C [7]. The main advantages of condensation heating for soldering are lack of overheating [8] and lack of shadowing effect, which occurs due to larger components [4]. However, intensive heat transfer during VPS and the hermetically closed process zone can also cause soldering failures like voiding, paste sputtering, tombstone failures [9], since the heat transfer coefficient of Galden vapour can be 2–3 times higher than the heat transfer coefficient of gas streams in a convection oven [10]. Up to now, only limited researchers have dealt with thermal aspects of the process. Leicht et al. showed that the heat transfer coefficient of vapour can be decreased with the application of non-saturated vapour
614
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
SMT PROCESS
a.)
STANDARD VPS PROCESS
Immersion
d.)
Assembled SMT PCB Vapour Galden
Printed Circuit Board (PCB) prepared for assembly
b.)
Stencil foil
Heaters
e.)
Paste deposit
Solder paste printing with stencil technology
Lift up
c.) Placing machine Auto component placement (assembled SMT PCB)
f.)
Dwelling in saturated vapour (heating)
Removal PCB, cooling
Fig. 1. The SMT and the basic VPS process [6].
[11]. Dumitru et al. investigated the effect of heating of VPS process on the mechanical characteristics of PCBs [12]. Cucu et al. studied the mechanical stability of the solder joint prepared by VPS process from low melting point alloys [13]. Krammer showed that a proper VPS thermal profile can increase the reliability of solder joints [14] since both the thickness and the shape of the intermetallic layer will differ from the conventional soldering technology. Synkiewicz et al. presented similar results about the general quality of the solder joints for thermo-generators [15]. Géczy et al. optimized temperature profiling methods for VPS process [16]; they also proved with pressure measurements that temperature saturation of the process zone occurs before vapour concentration saturation in time [17]. It was shown that condensate thickness changes considerably on the surface of the PCB which results in spatial heat transfer differences. In addition it was proven that the dense vapour space takes place considerably (20%) in the heating of the assembly [18,19]. The gas voids in the joints causing serious reliability problems since it decrease the joints life time and electrical conductivity. Krammer et al. proved that one of the key points to prevent of void formation is the appropriate amount of solder paste [20]. However, voids can even form during the stencil printing, which highly depends on the viscosity of the solder paste [21]. The issue of voiding was thoroughly investigated in the case of VPS and the problem was found even more serious than in the case of conventional soldering technologies [9]. Therefore, studies have started about the possible application of the low pressure for VPS system, where the voids are removed from the molten solder [22]. The suction process takes place while the alloy is above its melting point [22]. The pressure in the ‘‘vacuum chamber” is usually approximately 40–50 mbar during the suction part of the process [23]. However it was never studied that how does the decrease in vapour concentration affect the heat transport mechanism of the soldering process, and with this aspect, how does it affect the efficiency of the void removal and the forming of the solder joints microstructure. The question is relevant from the aspect of mass
manufacturing, while vacuum type ovens recently started to spread out in electronics assembly plants. 2. Numerical modelling of vacuum VPS process Investigating the VPS process by numerical simulations is necessary because of the closed environment (high temperature, dense vapour, etc.) in the soldering chamber. There were experimental studies about application of the temperature, pressure or optical probes in VPS ovens [16,24,25] but it was found that the introduction of the probes is complicated even in the case of simple VPS ovens. In the case of vacuum VPS process when the soldering chamber is hermetically closed, the measurement inside the soldering chamber is not possible with such probes which need to be connected to an outer device (like thermocouples, pressure sensors, borescopes, etc.). 2.1. Physical description of the model According to the preliminary calculations, the vapour concentration decreasing in the vacuum chamber causes a turbulent gas flowing [26], so the numerical calculations are based on the 3 dimensional Reynolds Averaged Navier-Stokes (RANS) equations. The RANS equations compute the average movement in a turbulent flow, while the effect of fluctuation is modelled by Reynold’s stress tensor which was related to the mean flow with a standard k-e turbulence model [27]. The decrease of the vapour concentration is initialized by a pressure drop generated by a vacuum pump. The governing equations of the vapour space are the following: the transient continuity equation for compressible Newtonian fluids is used since the local pressure change has effect on the density of the vapour:
@q @ui ¼0 þq @t @xi
ð1Þ
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
The RANS equation for compressible Newtonian fluids is used to close governing equations. Here it is formulated by the Einstein notation:
@ui @ui 1 @p @ 2 @uj ¼ gi þ ðt þ tT Þ 2Sij dij þ uj 3 @xj @t @xj q @xi @xj
ð2Þ
615
ing mass is determined by the heat flux which the condenser surface (in the case the suction pipe wall) can conduct away [29]:
@mc kðpÞ AðpÞ @T ðpÞ ¼ @t h @xj
ð10Þ
where A(p) is the area of the pipe wall [m2].
where the mean rate of strain tensor is:
Sij ¼
1 @ui @uj þ 2 @xj @xi
2.2. Numerical solution and parameters of the model
ð3Þ
where k is the turbulent kinetic energy:
The numerical conversion of the previously defined partial differential equations was done by Finite Difference Method (FDM) and was solved by explicit Forward Time Central Space (FTCS) algorithm in order to achieve high solving calculation speed and flexible implementation. The general numerical form of FTCS is the following:
@k @k ¼ þ uj @t @xj
@v @v @v 2 ¼0! þ þ FTCS @t @x @ 2 x
where the turbulent eddy viscosity is:
tT ¼ C l
k
2
ð4Þ
e 2 @uj 2 @ui 1 @ q @p tT 2Sij dij kdij tT 3 3 @xj 3 @xj q @xi @xi @ t @k eþ tþ T @xj rk @xj
ð5Þ
@e @e ¼ C e1 þ uj @t @xj
2 @uj 2 @ui 1 @ q @p tT 2Sij dij kdij tT 3 3 @xj 3 k @xj q @xi @xi e2 @ tT @ e tþ C e2 þ k @xj re @xj
e
ð6Þ The applied constants were the following (these are the generally used constants of the standard k-e model [28]):
C l ¼ 0:09;
rk ¼ 1; re ¼ 1:3; C e1 ¼ 1:44; C e2 ¼ 1:92;
The applied markings are the following: u is the velocity [m/s], m is the kinematic viscosity [m2/s] of the Galden vapour, p is the pressure [Pa], q is the concentration of the Galden vapour [kg/ m3], g is the gravity acceleration [m/s2] and dij is the Kronecker delta. The vapour pressure is calculated according to the general gas law:
p¼
amuH qRT amuGalden
ð7Þ
where amuH is the average molecular weight of hydrogen, amuGalden is the average molecular weight of Galden, R is the gas constant (8.314 J K1 mol1) and T is the temperature [K]. Turbulent transport of energy is also described by the averaging method:
@T @T k @ @T þ uj ¼ @t @xj qC s @xj @xj
ð8Þ
where k is the specific thermal conductivity [W/m K] and Cs is the specific heat capacity [J/kg K]. During the suctioning process, the vapour will condensate on the colder wall of the suction pipe. In this application, the formation of the condensate layer on the suction pipe wall and the presumed forming of dew point [19] at the end of the suction process are neglected. Therefore, the heat transfer by the condensation can be calculated only with the latent heat of the condensing mass [29]:
@T ðpÞ h @mc ¼ C ðpÞ @t @t
v
k iþ1
¼0
and e is the turbulent dissipation rate:
þ
v kþ1 v ki v kiþ1 uki1 i
ð9Þ
where T(p) is the pipe temperature [K], C(p) is the heat capacity of the pipe [J/K], h is the latent heat of the Galden [J/kg] and mc is the amount of condensing of Galden [kg]. The amount of the condens-
þ Dt 2 Dx k 2v i þ v ki1 Dx2
ð11Þ
where v is a general variable, t is the time [s], x is the space [m] and i and k are the indexing of the numerical grid and time steps respectively. The numerical model was built and solved in MATLAB software. A commercial vacuum VPS system was studied. The working principle of the oven is the following: until the melting of solder paste, it works like a simple VPS oven with a non-hermetic soldering chamber. After the melting of the solder alloy a cap closes the soldering chamber hermetically and forming a vacuum chamber with 325 325 85 mm dimensions. A suction pipe is located at the middle top of the closing cap with 13 mm inner and 24 mm outer diameter made from silicone rubber. The vacuum pump of the oven works with ‘‘volume separation” method which means that the pump separates the same volume of vapour out from the workspace in a given time. This implies constant decrease in the relative pressure between the suction pipe and the vacuum chamber. The vacuum pump decreases the pressure to 5 kPa in 5 s and the assembly spends further 15 s at this state to give enough time to the voids to leave the solder joints. (No more details or data about the working principle of the vacuum pump is known.). The possible effects of the heat transport changes during the vapour suctioning was studied on a FR4 Printed Circuit Board (PCB) with soldered resistors (with Al2O3 ceramic body). SAC305 type solder alloy was applied in 125 mm thickness (standard stencil thickness for solder paste printing) on the 17 mm thick (standard solder pad thickness) Cu solder pad. The surface of the PCB was 160 160 mm2 and 3 different thicknesses (0.5, 1 and 2 mm) were examined. The structure of the vacuum chamber with the assembly and the detailed structure of the assembly can be seen in Fig. 2. LS230 type Galden was used for the investigations. The physical parameters of the different materials can be seen in Table 1. The cell geometries and the time step was set by the Courant– Friedrichs–Lewy (CFL) condition, which has to be applied for all of the cells in the model:
Dt
ux uy uz 61 þ þ Dx Dy Dz
ð12Þ
According to the CFL condition, the soldering chamber was divided to 48,000 cells and the time step was set to 105 s. The assembly was divided to 2000 cells. The assembly model was embedded into the vacuum chamber model and it runs as a cosimulation (details about this modelling technique can be read in
616
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
Suction pipe Vacuum chamber Assembly
SAC305 solder
0.5mm Al2O3 resistor body
17µm thick Cu pad
0.5; 1; 2mm Fr4 board
Fig. 2. The structure of the soldering chamber with the assembly and the detailed structure of the assembly.
Table 1 Physical properties of the applied materials.
a b
Property/material
LS230
FR4
Cu
SAC305
Al2O3
Silicone rubber
Boiling point [°C] Density [kg/m3] Dynamic viscosityb [Ps s] Average molecular weight Specific thermal cond. [W/m k] Specific heat cap. [J/kg K] Latent heat [J/kg]
230 19.96a 2.13 105 1010 0.065 973 6.3 104
– 2100 – – 0.23 570 –
– 8960 – – 385 3850 –
– 7370 – – 58 2320 –
– 3690 – – 18 880 –
– 1420 – 0.35 1200 –
The saturated vapour concentration of the LS230 Galden. The dynamic viscosity of the LS230 Galden vapour was calculated by the method of Lee et al. in [30].
[19]). Due to the symmetry of the oven only a quarter of the whole geometry was calculated. At the start of the vapour suctioning, saturated vapour conditions were applied as an initial conditions. Saturated vapour condition was applied for the model as initial condition which means: q(0) = 19.96 kg/m3, T(0) = 230 °C, and the vapour is not moving before the suctioning, v(0)=0 m/s. Since the soldering chamber is not hermetically closed before the suctioning process the initial condition of the Galden vapour pressure is p(0) = 101 kPa. The initial temperature of the assembly is also 230 °C. Initial condition of the turbulent kinetic energy:
3 k ¼ ðU IÞ2 2
ð13Þ
where U is the mean flow velocity and I is the turbulence intensity which was chosen to 0.1 (the flow is supposed to be highly turbulent). Initial condition of the turbulent dissipation rate: 3
e ¼ Cl
k2 l
ð14Þ
where l is the turbulent length scale which is 0.1 m in this case. During the preliminary investigations it was found that during the suctioning process the soldering chambers suffers a heat loss on the suction pipe. The initial temperature of the suction pipe
and its change during the suctioning was measured with K-type thermocouples. It was found that initial temperature of the suction pipe is Tp(0) = 60 °C which increases 1.6 °C during the 5 s of suctioning. Fig. 3 shows a proper fitting between the calculated and measured temperatures of the suction pipe during the suctioning process. Boundary conditions of the model are the followings: the walls of the soldering chamber are adiabatic and the flow velocity tends to zero at the walls:
@T ¼ 0 @r r!wall
and
lim
v ¼0
r!wall
ð15Þ
3. Results and discussion The first step of the investigations was to determine the relative pressure decrease generated by the vacuum pump which is necessary to reach the 5 kPa pressure of the soldering chamber in 5 s. During the preliminary calculations with a laminar flow model it was found that this relative pressure decrease is 1.22%p, however the preliminary calculations also showed that according to the Reynolds numbers the flow has to be highly turbulent [26]. The
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
Fig. 3. Calculated and measured temperature of the suction pipe during the suctioning.
turbulence decreases the average flow velocity, therefore it was supposed that in reality the relative pressure decrease of the vac-
617
uum pump has to be higher than 1.22%p. The approximation started from 1.22%p in 0.01 steps. Fig. 4a) shows the pressure and the concentration decrease in the soldering chamber in the case of different vacuum pump settings. The results shows that the -3.4%p vacuum pump setting approximates the real operation, in this case the vapour concentration decreases to 1.98 kg/m3. The ratios of the pressure and concentration decreases are not the same since the soldering chamber suffers a heat loss during the suctioning process. This results in that the pressure decreases to 5% while the vapour concentration only to 10%. The flow velocity analyses showed that at the beginning of the suctioning process (between 0 and 0.075 s) the movement of the vapour is highly fluctuating in the soldering chamber (Fig. 4b and c). This is caused by the fluctuation of the pressure. After 0.075 s till the end of the suctioning process the flow velocities remains constant. At the beginning of the suctioning (0.01 s) the flow velocity reaches 28 m/s at the entrance of the suction pipe, and after a high drop down, it is stabilizing around 14 m/s (Fig. 4b). There are high flow velocity differences diagonally in the soldering chamber, the vapour almost stands at the corners and the flow velocity is still only 2 m/s at the half of the diagonal space (Fig. 4b). Towards the vertical direction (under the suction pipe) the velocity differences are smaller, the flow velocity is 9 m/s at the bottom of the soldering chamber (Fig. 4c). It can
Fig. 4. Vapour space parameters during the suctioning process: (a) decrease of the pressure and vapour concentration in the case of different vacuum pump settings; (b) flow velocities diagonally in the soldering chamber (3.4%p) between 0–0.2 s; (c) flow velocities vertically in the soldering chamber (3.4%p) between 0–0.2 s.
618
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
the top and the bottom level of the soldering chamber the vapour concentration difference is: 1.5 kg/m3 (8.3%) at 0.1 s and 0.1 kg/ m3 (7.1%) at 5 s. According to the energy transport calculations, the heat loss on the suction pipe causes considerable heat loss in the soldering chamber during the suctioning. Fig. 6 shows the temperature changes during the whole void removal process (20 s) in the case of different settings. During the suctioning (0–5 s) the vapour temperature decreases to 80 °C (Fig. 6b). The vapour temperature differences are low during the suctioning (between 2–4 °C). However during the rest of the void removal process (between 5–20 s), the vapour nearby the assembly is heated back (to 130– 145 °C) by the hotter assembly, but this heating effect on the whole chamber is moderate. The observed heat loss and the vapour concentration differences effect on the heat transport processes of the soldering. It is possible that the solder joints can be solidified much earlier than the end of the void removal and the voids cannot leave the solder joints. This effect was studied on solder joints (Fig 2) placed at different positions on different thick PCBs (Fig. 6a). Fig. 6c) shows the minimum and maximum solder joint temperatures on PCB with different thickness during the void removal. During the suctioning the temperature of the solder joints decreases considerably with 8–14 °C. In the cooling of the solder joints the thickness of the PCB is important since this determines the total heat of the assembly. The temperature decrease is the most serious in the middle of the PCB (point 9), the heat loss is less towards the corners of the
be concluded that the suctioning moves the vapour the most in the middle space of the soldering chamber. In the next step the changes of the vapour concentration was studied during the suctioning process at different levels of the soldering chamber. Fig. 5 shows the vapour concentration at the bottom, at the middle and at the top of the soldering chamber. At the beginning of the suctioning process (0–0.03 s), the fluctuation of the flow velocity changes the distribution of the vapour concentration (Fig. 5a and b). Up to 0.1 s the distribution of the vapour concentration is stabilizing and remaining similar till the end of the suctioning (Fig. 5c and d). It was found that after the distribution of the vapour concentration stabilized, considerable concentration gradient can form, mainly horizontally, between the suction pipe and the corners of the vacuum chamber (Fig. 5c). The maximum difference is 2.25 kg/m3 (12%) at the top level of the soldering chamber at 0.1 s. During the suctioning this differences is decreasing slightly, at the end of the process (Fig. 5d), it is 0.15 kg/m3 (10%). Vertically, along the height of the soldering chamber, smaller vapour concentration differences were found than horizontally. Under the suction pipe between the top and the bottom level of the soldering chamber the vapour concentration difference is: 1.5 kg/m3 (8.3%) at 0.1 s and 0.1 kg/m3 (7.1%) at 5 s. During the suctioning this differences is decreasing slightly, at the end of the process (Fig. 5d), it is 0.15 kg/m3 (10%). Vertically, along the height of the soldering chamber, smaller vapour concentration differences were found than horizontally. Under the suction pipe between
b) Vapour concentration [kg/m3]
a) Vapour concentration [kg/m3]
20
Bottom
20.5
Bottom
19.5
20
19
19.5
Middle
19
Middle
18.5 18
18.5
Top Top
18 18
18 12
12
Cham
6
6
ber d
im. [c
m]
0 0
im.
ber d
Cham
17.5 18
[cm]
18 12
Cham12 ber d im
6
. [cm]
6 0 0
. [cm]
im
ber d
Cham
d) Vapour concentration [kg/m3]
c) Vapour concentration [kg/m3]
1.5
19
Bottom
Bottom
18.5
1.45
18
1.4
17.5
1.35
Middle
17
Middle
1.3
Top
Top
16.5 18
18 12
12
Cham
ber d
6
6
im. [c
m]
0 0
Cham
ber d
m] im. [c
1.25 18
18 12
12
Cham
ber dim
6
. [cm]
6
0 0
. [cm]
er dim
b Cham
Fig. 5. Vapour concentration at different levels of the soldering chamber (quarter sections, 3.4%p): (a) at 0.015 s; (b) at 0.03 s; (c) at 0.1 s and (d) at 5 s.
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
619
Fig. 6. Temperature changes in the soldering chamber during the void removal: (a) component positions on the PCB; (b) temperature changes of the vapour; (c) minimum and maximum solder joint temperatures in the case of different PCB thicknesses; (d) minimum solder joint temperatures in the case of different oven settings.
PCB (point 1). This effect is caused by the characteristic of the suctioning, namely that the vapour concentration (Fig. 5c and d) as well as the vapour temperature (Fig. 6a) is the lowest at the middle of the soldering chamber where the assembly is usually positioned. The solidus temperature of the lead free solder alloy is around 219 °C, according to the calculations, if the PCB is thinner than 2 mm, the solder joints can be solidified (or get very close to the solidification) to the end of the suctioning. Although after the suctioning the temperature of the solder joints increases a bit with 2 °C, it is caused by the heating effect of the assembly (the PCB and the ceramic body of the components remain at higher temperature than the solder joints) and the stop of the convection. This can cause that the solidified joints melt again, but the time of the void removal will be still shorter than it has to be. Different oven settings were studied in order to decrease the heat loss during the suctioning process. Fig. 6d) shows the lowest solder joint temperatures on a 1.5 mm thick PCB in the case of different oven settings. The intensity of the vapour concentration decrease has significant effect on the heat loss since it determines the flow velocity and via this the convection. If the relative pressure decrease of the vacuum pump is only 2.5%p (instead of 3.4%p) the positive effect is already visible (Fig. 6d), the temperature of the solder joints remains over 221.5 °C. However the 221.5 °C is still very close to the liquids point and with this vacuum
pump setting the pressure remains 20 kPa in the soldering chamber after 5 s suctioning (Fig. 4a) which ensures less effective void removal than the 5 kPa in the case of 3.4 %p. Since the heat loss is happening on the suctions pipe, the heating of the suction pipe would be an evident solution. If the temperature of the suction pipe would be elevated from 60 to 120 °C, considerable improvement could be reached. The solder joints temperature remains over 223 °C (Fig. 6d) without losing the suction power of the low pressure in the soldering chamber. The thermal diffusivity of the vacuum pipe was also examined, but the thickness increase of the vacuum pipe wall has got only a minor effect on the heat loss. In the last step the calculations were done with a heat transfer fluid with higher boiling point (Galden HS240, boiling point 240 °C). This change showed the best solution, the temperature of the solder joints remains over 229 °C during the whole void removal process (Fig. 6d). 4. Conclusions The heat and mass transport processes was investigated by numerical simulations in a vacuum type VPS oven during the vapour suctioning process. For this purpose a three-dimensional numerical gas flow model was developed which based on the RANS equations with the standard k-e turbulence method. It was found
620
B. Illés et al. / International Journal of Heat and Mass Transfer 114 (2017) 613–620
that during the vapour suctioning, high flow velocity differences can form in the soldering chamber which causes considerable vapour concentration gradient, mainly horizontally, between the suction pipe and the corners of the vacuum chamber. The oven suffers a heat loss on the suction pipe which causes the drop of the solder joints temperature with 8–14 °C (determined by the position of the given joint on the PCB and the thickness of the PCB). The combined effect of the vapour concentration gradient and the temperature drop can cause the following phenomenon: the solder joints will solidify during the vapour suctioning process before the voids could be removed from the solder joints. This side effect inhibits the further void removal from the solder joints and this has negative effect on the formation, and the overall quality and reliability aspects of the solder joints microstructure. In order to ensure effective void removal, the vacuum VPS technology has to be set/modify carefully (e.g. heating of the suction pipe) mainly in the function of the heat capacity of the soldered assembly. Nevertheless, in the case of common lead-free alloys, the application of heat transfer fluid with the highest suggested boiling point (240 °C) and PCBs having thicknesses 1.5 mm or thicker are suggested for vacuum VPS technology. Conflict of interest The authors declared that there is no conflict of interest. Acknowledgement This research has been partially supported by the ‘‘ÚNKP” program of the Hungarian Government. References [1] C. Naldi, E. Zanchini, Dynamic simulation during summer of a reversible multifunction heat pump with condensation-heat recovery, App. Therm. Eng. 116 (2017) 126–133. [2] J. Kaew-On, N. Naphattharanun, R. Binmud, S. Wongwises, Condensation heat transfer characteristics of R134a flowing inside mini circular and flattened tubes, Int. J. Heat Mass Trans. 102 (2016) 86–97. [3] F.L. Chang, Y.M. Hung, Dielectric liquid pumping flow in optimally operated micro heat pipes, Int. J. Heat Mass Trans. 108/A (2017) 257–270. [4] B. Illés, I. Bakó, Numerical study of the gas flow velocity space in convection reflow oven, Int. J. Heat Mass Transf. 70 (2014), 195-191. [5] B. Kovács, A. Géczy, G. Horváth, I. Hajdu, L. Gál, Advances in Producing Functional Circuits on Biodegradable PCBs, Period. Polytech. Elec. Comp. Sci. 60 (4) (2016) 223–231. [6] B. Illés, A. Géczy, Investigating the heat transfer on the top side of inclined printed circuit boards during vapour phase soldering, App. Therm. Eng. 103 (2016) 1398–1407. [7] R.P. Nielsen, R. Valsecchi, M. Strandgaard, M. Maschietti, Experimental study on fluid phase equilibria of hydroxyl-terminated perfluoropolyether oligomers and supercritical carbon dioxide, J. Supercrit. Fluids 101 (2015) 124–130. [8] J. Huang, J. Zhang, L. Wang, Review of vapor condensation heat and mass transfer in the presence of non-condensable gas, App. Therm. Eng. 89 (2015) 469–484.
[9] B.K. Synkiewicz, A. Skwarek, K. Witek, Voids investigation in solder joints performed with vapour phase soldering (VPS), Solder. Surf. Mount. Tech. 26 (1) (2014) 8–11. [10] H.K. Dawood, H.A. Mohammed, N.A.C. Sidik, K.M. Munisamy, M.A. Wahid, Forced, natural and mixed-convection heat transfer and fluid flow in annulus: a review, Inter. Comm. Heat Mass Transf. 62 (2015) 45–57. [11] H. Leicht, A. Thumm, Today’s Vapor Phase Soldering - An Optimized Reflow Technology for Lead Free Soldering, in: Proceed. of Surf. Mount Tech. Assoc. Intern. Conf., Orlando, USA, paper No. 45, 2008. [12] G.I. Dumitru, A. Tudor, G. Chisiu, I. Plotog, Mechanical characteristics of electronic printed circuit obtained by the vapour phase soldering process, App. Mech. Mater. 658 (2014) 71–76. [13] T.C. Cucu, I. Plotog, M. Branzei, Mechanical tests regarding low-temperature lead-free solder pastes application in automotive electronics, in: Proc. of 20th International Symposium for Design and Technology in Electronic Packaging (SIITME), Bucharest, Romania, 2014, pp. 63–68. [14] O. Krammer, Comparing the reliability and intermetallic layer of solder joints prepared with infrared and vapour phase soldering, Solder. Surf. Mount Technol. 26 (2014) 214–222. [15] B. Synkiewicz, A. Skwarek, K. Witek, Vapour phase soldering used for quality improvement of semiconductor thermogenerators (TEGs) assembly, Mater. Sci. Semicond. Process. 38 (2015) 346–351. [16] A. Géczy, B. Kvanduk, B. Illés, G. Harsányi, Comparative study on proper thermocouple attachment for vapour phase soldering profiling, Solder. Surf. Mount Technol. 28 (2016) 7–12. [17] A. Géczy, B. Illés, Z. Péter, Z. Illyefalvi-Vitéz, Characterization of vapour phase soldering process zone with pressure measurements, Solder. Surf. Mount Technol. 25 (2) (2013) 99–106. [18] B. Illés, A. Géczy, Investigating the dynamic changes of the vapour concentration in a Vapour Phase Soldering oven by simplified condensation modelling, App. Therm. Eng. 59 (2013) 94–100. [19] B. Illés, A. Géczy, Numerical simulation of condensate layer formation during vapour phase soldering, App. Therm. Eng. 70 (2014) 421–429. [20] O. Krammer, B. Varga, K. Dušek, New method for determining correction factors for pin-in-paste solder volumes, Solder. Surf. Mount Technol. 29 (1) (2017) 2–9. [21] O. Krammer, B. Varga, D. Busek, Investigating the effect of solder paste viscosity change on the pressure during stencil printing, in: Proc. of 22nd International Symposium for Design and Technology in Electronic Packaging (SIITME), Oradea, Romania, 2016, pp. 36–39. [22] C. Zabel, A. Duck, Vapour phase vacuum-soldering. A new process technology opens tremendous production capabilities when reflow-soldering, SMTA International Conference, San Diego, California, October 4–8, 2009. [23] C. Zabel, A. Duck, Vapour phase vacuum-soldering. A new process technology opens tremendous production capabilities when reflow-soldering, SMTA International Conference, San Diego, California, October 4–8, 2009. [24] A. Géczy, D. Nagy, B. Illés, L. Fazekas, O. Krammer, D. Busek, Studying heat transfer on inclined printed circuit boards during vapour phase soldering, Solder. Surf. Mount. Tech. 29 (1) (2017) 34–41. [25] L. Livovsky, A. Pietrikova, Real-time profiling of reflow process in VPS chamber, Solder. Surf. Mount. Tech. 29 (1) (2017) 42–48. [26] B. Illés, A. Skwarek, A. Géczy, K. Witek, Numerical Study of Vapour Concentration Reduction in a Vacuum Vapour Phase Soldering Oven, in: Proc. of 40th International Spring Seminar on Electronics Technology, Bulgaria, Sofia, 2017, in press. [27] Y.-Y. Bae, E.-S. Kim, M. Kim, Assessment of low-Reynolds number k-e turbulence models against highly buoyant flows, Int. J. Heat Mass Trans. 108/A (2017) 529–536. [28] H.K. Versteeg, W. Malalasekera, in: An Introduction to Computational Fluid Dynamics: The Finite Volume Method, Chapter 3: Turbulence and its modelling, second ed., Harlow, Pearson Education, 2007, pp. 42–65. [29] B. Illés, A. Géczy, A. Skwarek, D. Busek, Effects of substrate thermal properties on the heat transfer coefficient of vapour phase soldering, Int. J. Heat Mass Trans. 101 (2016) 69–75. [30] A.L. Lee, M.H. Gonzalez, B.E. Eakin, The viscosity of natural gases, J. Pet Technol. 18 (8) (1966) 997–1000.